# Deployed

Enterprise AI deployments in production, published by AIM Media House.

## Anthropic's Mythos Found the Bugs. Bank of America Hired 1,000 Contractors to Fix Them.

Date: 2026-10-23T12:51:38.480Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/anthropics-mythos-found-the-bugs-bank-of-america-hired-1000-contractors-to-fix-them

Bank of America CEO Brian Moynihan said on October 14, 2026 that the bank had to scale up its patching work after Anthropic’s Mythos model “started finding bugs in the system.” “We put on 1,000 contractors to help us do the patching work in the second quarter,” he told CNBC . The contractors were added on top of about 3,000 people already working in that area. According to Moynihan, the flaws were in code used by both the bank and its outside vendors. On the bank's Q3 earnings call the same day, Moynihan put a figure on the combined effort. Responding to Wells Fargo analyst Mike Mayo, Moynihan said the bank spent “a couple hundred million plus” in the third quarter on two things: remediation for its anti-money-laundering orders and “remediation for the Mythos implementation.” He did not break out how much of that went to Mythos. He said the bank expects "that the Mythos remediation costs run down because we added 1,000 extra staff to work on that, not in the headcount but in consultants." The Same Quarter, Both Sides of the AI Ledger The disclosure came on the same call where Moynihan laid out the productivity case for AI. Bank of America has about 130 AI projects implemented , and Moynihan said "all of them have a payback." AI-related benefits are running at $250–300 million this year and are expected to reach $700 million in 2027. While the remediation workforce grew, the bank's own headcount shrank. CFO Alastair Borthwick said headcount fell by nearly 3,500 year over year "as we automate processes and simplify work across the company." The 1,000 people brought in for Mythos remediation do not appear in that figure, since Moynihan said they were added "not in the headcount but in consultants." The remediation figure sits alongside those numbers. Bank of America's AI benefits for the full year and its combined Mythos and AML remediation spend for a single quarter are in the same range: hundreds of millions of dollars. One is a projected return on deploying AI. The other is the cost of responding to what an AI model found. The AML half of the figure has its own context. Bank of America took a $1.2 billion regulatory settlement charge in Q3 related to a government anti-money-laundering matter, which Moynihan said has now been resolved. CFO Alastair Borthwick told analysts the bank has "ramped

## Brown & Brown Launches COMPASS to Bring AI Into Insurance Buying Decisions

Date: 2026-10-23T12:14:53.710Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/brown-brown-launches-compass-to-bring-ai-into-insurance-buying-decisions

Brown & Brown introduced COMPASS on October 22, 2026, an intelligence initiative that combines the brokerage's proprietary data, analytics, insurance expertise, and AI capabilities. The platform is designed to support risk assessment, insurance purchasing, and portfolio optimization across Property & Casualty, Employee Benefits, and Personal Lines, according to the company. The rollout starts internally. Brown & Brown expects COMPASS-powered knowledge assistants and business intelligence tools to reach thousands of employees by the end of 2026, beginning in the U.S. An expanded suite of customer solutions across its three main segments is planned for early 2027. "COMPASS brings together capabilities that have traditionally operated separately, giving our teams a more complete view of a customer's risks and opportunities and helping turn fragmented information into relevant insight," said Steve Hearn, president of Brown & Brown Retail. "That broader context can support more informed recommendations, better decisions and stronger outcomes across the insurance value chain." Powell Brown, president and CEO, said the platform is intended to connect institutional knowledge across the organization. "When we help our customers make more informed decisions, everybody wins," he said. Brown & Brown cited two existing analytics capabilities as examples of the approach. In Employee Benefits, the company said customers using its Health Plan Intelligence tools experienced an average five-year cost trend below 4%, compared with an industry and market norm of approximately 7% to 9%. In Property & Casualty, the company said COMPASS analytics helped a Fortune 50 customer restructure its insurance program, cutting premium spending by 65% while substantially reducing potential risk volatility. Brown & Brown did not name the customer. For Personal Lines, COMPASS will support digital quoting and servicing. Offerings for ultra-high-net-worth customers and single-family offices will include analytics to evaluate risk transfer and optimize insurance portfolios. The initiative builds on Brown & Brown's Digital Innovation and Insights Group, along with the company's collaborations with Anthropic, McKinsey, and Accenture.

## “Knowing an Agent Is Deployed Isn’t Enough,” Says CrowdStrike’s Field CTO

Date: 2026-10-23T11:03:12.812Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/knowing-an-agent-is-deployed-isnt-enough-says-crowdstrikes-field-cto

AI agents are already behaving in ways enterprises did not plan for, according to Fabio Fratucello, Field CTO World Wide at CrowdStrike. "Unexpected agent behavior is already occurring in real enterprise environments, and as models become more capable, enterprises require stronger controls," Fratucello told AIM Media House during an exclusive interview. He traced the risk to how agents are built. "Agents are goal-seeking by design," he said. "They can execute exactly what was requested, encounter an obstacle, find an alternative path, and take unconventional, and potentially unauthorised actions." Fratucello pointed to sandbox escapes documented by frontier AI labs as evidence of what increasingly capable agents can do, adding that “these scenarios extend beyond testing environments.” In production, Fratucello said, "prompt injection remains the most critical attack pattern because it can manipulate agents into taking actions outside their intended outcome." He listed the other common risks as "credential abuse, excessive permissions, malicious tool use, data exfiltration, and compromised agents executing malicious code." The Endpoint as the Control Point Fratucello said the most dangerous assumption he sees is treating a successful login as the end of the security check. "Treating authentication as the end of the security decision creates significant risk. Agents are being deployed with broad, standing permissions inherited from humans, when they require continuous control instead, he said," He argued that the endpoint is where that control has to sit. "The endpoint serves as the optimal control point because agents execute there; they use credentials, access data, invoke tools, and take action. Network and application controls remain important, but are limited in visibility and lack enforcement capabilities where it matters the most." Enterprises, he said, "must observe the complete execution chain at runtime, connecting what the agent was asked to do with what it actually accomplished, and enforcing policy and technical control before that action creates impact." CrowdStrike's Falcon Guardian, he said, "delivers the comprehensive visibility and enforcement protection enterprises need to secure AI agents during execution." CrowdStrike unveiled Falcon Guardian, an AI Detection and Response solution, at its Fal.Con 2026 conference on September 1, 2026. Falcon Guardian includes AI Agent Discovery and Inventory, which discovers known and shadow AI agents across Windows and macOS. Fratucello described shadow AI as "a massive visibility challenge," saying "organisations deploy agents faster than they can govern them." He said unapproved agents "extend beyond unsanctioned enterprise agents to include developer and coding agents, personal

## Visa’s AI Agents Can Now Pay Business Invoices

Date: 2026-10-23T09:43:11.950Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/visas-ai-agents-can-now-pay-business-invoices

Visa announced three capabilities across commercial card growth, virtual card security, and agentic commerce on October 21, 2026, designed to help financial institutions, fintechs, and businesses automate payment processes while maintaining "the trust, visibility and control those transactions require." The most direct move into AI agents is a live pilot with BILL and Billtrust that enables trusted AI agents to initiate and complete approved business invoice payments through supplier portals. The pilot is processing end-to-end transactions with participating BILL customers and Billtrust suppliers. "Businesses are moving quickly to automate the work around payments, from identifying supplier opportunities to approving invoices and executing transactions," said Darren Parslow, global head of Visa Commercial Solutions. He said the capabilities are designed to help businesses do that "whether they are growing commercial card acceptance, helping protect virtual-card payments from fraud, or testing how trusted AI agents can act on their behalf." The pilot draws on Visa Intelligent Commerce and Visa's Trusted Agent Protocol to establish the agent's identity, confirm that it is authorized to act, and ensure each transaction follows business-defined controls. Instead of relying on shared supplier-portal login credentials, it uses tokenized payment credentials and connects buyer-side invoice approval with supplier-side payment acceptance., according to the company Visa said the setup gives businesses visibility into who is acting, what the agent is permitted to do, and which transactions have been approved. SESA 2.0, the next generation of Visa's Supplier Enablement Spend Analytics, combines AI-driven supplier scoring, Visa commercial payments data, buyer benchmarks, and business-value modeling. According to the company, it helps financial institutions identify suppliers that may be more likely to accept commercial card payments and estimate potential commercial card growth. Since its launch, SESA has helped financial institutions uncover more than $600 billion in potential cardable spend, Visa said. Visa Commercial Fraud Solutions covers three stages of a commercial payment. Supplier Validation helps verify payments reach the intended supplier. Virtual Authentication verifies commercial virtual card users through 3-D Secure authentication . Payment Controls let businesses set transaction-level limits on spend, merchant category, location, and timing. Payment Controls are available now to eligible issuing financial institutions in supported countries and territories. Supplier Validation and Virtual Authentication are in pilot and will be generally available in 2027. Visa cited Juniper Research projections that global B2B virtual card payment volume will reach $14.6 trillion by 2029, representing 83% of the global virtual card market. As

## Walmart Is Opening Its Store to AI Agents. Customers Set the Spending Limits.

Date: 2026-10-22T12:01:05.425Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/walmart-is-opening-its-store-to-ai-agents-customers-set-the-spending-limits

Walmart announced on October 21, 2026 a set of tools that will let AI agents shop and complete purchases on behalf of customers. The company is introducing Agent Profiles, the Walmart Agent Gateway, and a set of terms for AI agents, which together will let customers shop with Walmart and Sam's Club "through the apps and agents they choose, with clear permissions, controls and protections built in." Walmart already serves customers through conversational commerce integrations with Meta's Muse, OpenAI's ChatGPT, Google's Gemini, and others. Reuters noted that the move contrasts with Amazon, which has sought to limit third-party AI shopping agents' access to its platform. Today, when a customer uses an AI agent on Walmart, the agent signs in using the customer's account, and all purchases appear together without differentiation. Agent Profiles change that. Each authorized agent receives a dedicated profile within the customer's account, so its purchases can be viewed separately. Customers can require approval for every purchase or allow the agent to buy within a spending limit they set, and can delete an Agent Profile at any time. Walmart compared the setup to "adding someone as a cardholder on your credit card." The company said prices shown to an agent will match those shown to a human on Walmart.com. "We price products, not people," Walmart said. The Walmart Agent Gateway addresses how agents shop. Agents currently navigate Walmart's site through a web browser, following steps designed for people, which Walmart said "can be slow and error-prone." The gateway gives agents a faster, more reliable way to shop using MCP, and will begin with a beta for companies and developers. It will also help agents find a customer's usual items, such as a preferred brand of milk, without sharing the customer's broader shopping history with the agent. Walmart's terms set out what agents can and cannot do. Customers must approve agent purchases by default, and Walmart may require additional approvals even within an agent's spending limit. Agents must not take actions beyond what the shopper intends, cannot add or change payment methods, and can access only the information needed for the task. Walmart's returns process applies, including if an agent does not act as intended. Agents that break the terms will be blocked. The tools will be available to all Walmart account holders and will not require a Walmart+ membership. Mizuho analyst David Bellinger said in a note that

## AEP's new AI security chief gets the strategy, not the veto

Date: 2026-10-22T10:57:50.165Z
Author: Deployed
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/aeps-new-ai-security-chief-gets-the-strategy-not-the-veto

American Electric Power serves 5.6 million customers in 11 states and runs 40,000 miles of transmission line, the largest network in the country. On 21 October it posted two senior security jobs on the same day, with the same pay range, both on site in Columbus or Austin: a Managing Director for OT Security and a Managing Director for AI Security. The OT job is what you would expect from a utility: substations, SCADA, protective relays, nuclear plant controls. The AI job is the unusual one. It describes a “Cyber AI” program inside the cybersecurity organisation, with two halves: protecting the company’s own AI, and using AI to run security better. The posting calls them AI Security Operations and AI-enabled cybersecurity products and engineering. On paper it is a big job. The holder sets the multi-year Cyber AI strategy, builds the teams, guides the investment and “represents cybersecurity in enterprise AI decisions.” The pay range is $188,524 to $245,080. Then comes a section most job ads do not have. It is headed “Role Boundaries”, and it lists what the new managing director does not control. Ten things the AI security chief cannot decide The posting says the role “does not replace established enterprise authority” for architecture approval, cloud and infrastructure operations, enterprise incident command, data governance, privacy, legal interpretation, compliance, model development, business product ownership or enterprise risk acceptance. Those stay with their existing owners. The job’s tools are strategy, portfolio, reporting and escalation. That last item matters most. If a business unit wants to deploy an AI agent that the Cyber AI team considers unsafe, the managing director can escalate it. Accepting or rejecting the risk belongs to someone else. Our view: for a company that keeps the lights on for 5.6 million customers, that is the wrong design. As AI spreads through a utility’s operations, it will end up close to the systems that keep the grid running. The person accountable for AI security should be able to stop a deployment, not just write a memo about it. The OT job gets the sharper tools Read the two postings side by side and the difference is plain. The OT Security director is asked to establish “delegated decision authority” across business units and vendors, and to keep the technical depth to challenge high-consequence engineering conclusions. The AI Security director gets “decision boundaries” and “escalation paths”. The OT side has

## AI Debt Drove Over Half of Moody's Issuance Growth This Year

Date: 2026-10-22T10:55:35.948Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/ai-debt-drove-over-half-of-moodys-issuance-growth-this-year

“AI-related issuance has accounted for a little over half of total issuance growth” year to date, Moody’s CEO Rob Fauber said on the company’s Q3 2026 earnings call on October 21, 2026, adding that its share of revenue growth is “closer to 40%.” The growth is concentrated in digital infrastructure. Fauber said digital infrastructure issuance rose approximately fourfold from Q3 2025. CFO Noémie Heuland said investment-grade revenue rose 18% on 16% issuance growth in the quarter, supported by AI infrastructure and hyperscaler financing. To rate that activity, Moody's has introduced a cross-sector digital infrastructure methodology covering AI infrastructure build-outs, according to the company. Fauber cited Moody's rating analysts' estimate of "something like $800 billion of CapEx for 2026," which they expect to grow to "north of $1 trillion" in 2027. On the near term, Fauber said Moody's has "assumed that the hyperscalers have done most of their financing for the year," meaning any further hyperscaler issuance in the fourth quarter would be upside to its outlook. The company expects Q4 issuance to fall by a mid-teens percentage year over year and maintained its guidance for high-single-digit revenue growth in Moody's Investors Service for the full year. Moody's Investors Service revenue was $1.09 billion in Q3, down 1%, with issuance down 2%. Moody's also raised its full-year outlook for first-time mandates to 900–950. On the Moody's Analytics side, Fauber said customers now use the company's AI in three ways: "AI that's embedded in our own applications," through agents and workflows, and through direct MCP and API access. MCP users "more than quadrupled" and usage rose more than tenfold over the last three months. "The number of customers paying for MCP access has more than doubled in this last quarter," Fauber said, adding that it is "still off of a small base." Analysts using the MCP are "reaching 3 times more content on a working day." Fauber said one large bank with 8,000 users on a single AI surface wants Moody's content on that surface. Clear Street analyst Owen Lau asked how Moody's will avoid cannibalizing its own platform while distributing data through AI models and MCP. Fauber said pricing for MCP content may include "elements of consumption-based pricing," with more detail expected in early 2027. Fauber also said Moody's AI-enabled lending suite is "one of our fastest-growing offerings," with lending ARR up 12%, and described its Banking Decision Intelligence product as

## Whoever runs the GPUs should run the AI strategy

Date: 2026-10-22T09:53:53.063Z
Author: Deployed
Desk: Pharma
URL: https://aimmediahouse.com/pharma/whoever-runs-the-gpus-should-run-the-ai-strategy

Every academic health system now has an AI story. Most of them are about tools: an ambient scribe here, an inbox assistant there, a pilot with a model vendor announced at a conference. Very few are about the thing underneath all of them, which is computing capacity, and who decides how it is bought, shared and paid for. UT San Antonio has just put that question at the centre of its org chart. On 19 October its senior vice president and CIO, Michael Schnabel, announced on LinkedIn a “newly established leadership role”: a Senior Director of Research Computing, building one function across the university’s colleges and its health system. The title undersells it. The posting calls the role the institution’s “principal advisor on artificial intelligence”, responsible for enterprise AI strategy, governance and adoption “across research, clinical, educational, and administrative functions.” The same person will run high-performance computing, GPU clusters, secure enclaves for regulated data, cloud landing zones on AWS, Azure, Google Cloud and Oracle, and the link to the Texas Advanced Computing Center. The person who owns the GPUs will also own the AI strategy. That is the right call, and much of academic medicine has it the other way round. Why compute, and not an AI office A common pattern in health systems has been to appoint an AI leader whose remit is use cases and governance committees, set apart from the people who run the infrastructure. That produces an office that can approve a tool but cannot say what it will cost at scale, where the data will run, or whether the cluster has room for the next grant. Schnabel has been blunt about the cost side. In a LinkedIn post last month on “tokenomics” in Epic’s AI pricing, he warned that health systems are moving from predictable licence fees to consumption-based costs, and that a cheap-looking AI feature “can create a budget problem several years down the road.” His prescription was dashboards and thresholds: what are we spending, where, and what are we getting back. Michael Schnabel, Senior Vice President and Chief Information Officer, UT San Antonio. Photo: UT Health San Antonio “The technology is moving faster than our traditional budgeting and governance models.” Michael Schnabel, on LinkedIn, September 2026 You cannot answer those questions from a committee room. You answer them where the workloads run. Putting AI strategy inside research computing ties the policy to the

## AstraZeneca Bets on Oracle to Find Trial Patients Faster

Date: 2026-10-22T09:35:17.245Z
Author: Amit Raja Naik
Desk: Pharma
URL: https://aimmediahouse.com/pharma/astrazeneca-bets-on-oracle-to-find-trial-patients-faster

Oracle and AstraZeneca recently announced an agreement to use AI and real-world data to help clinicians identify patients for clinical trials, improve recruitment and bring research findings into everyday care. “With this collaboration between AstraZeneca and Oracle, we can use real-world data to identify and match qualified patients with relevant clinical trials while cutting costs and bringing insights into care faster,” wrote Seema Verma , executive vice president and general manager of Oracle Health and Life Sciences, on X. The non-binding agreement combines AstraZeneca’s scientific and therapeutic expertise with Oracle’s health data, cloud, AI and clinical software. Financial terms were not disclosed, and the companies have not announced a cost-savings target. “Combining AstraZeneca’s world class science with Oracle’s deep data and technology expertise, we can help clinicians identify patients sooner, optimize clinical trials, and generate evidence continuously across the lifecycle of a medicine,” said AstraZeneca CEO Pascal Soriot. The first planned program will focus on hypertension care and clinical research in Abu Dhabi, working with healthcare company M42. It aims to identify patients with uncontrolled hypertension, support guideline-directed care and improve trial recruitment. AstraZeneca identified the US, Middle East and Europe as regions for the broader collaboration. Participating US health systems and a US rollout date have not been specified. Why Oracle’s Data Matters In September, Oracle expanded Life Sciences Data Intelligence with AI capabilities that allow researchers to explore patient groups using natural language and automate multistep analyses. The platform supports trial recruitment, site selection and evidence generation. Oracle says its real-world data resource contains more than 122 million de-identified patient records, although the partners have not specified which datasets they will use. “Fragmented data and disconnected workflows continue to slow the path to discovery,” Verma said in the September announcement. A separate 2024 NIH-led TrialGPT study found a 42.6% reduction in patient-screening time in a pilot user evaluation. Oracle and AstraZeneca have not reported screening or enrollment results from their collaboration. AstraZeneca’s Broader AI Push Across Drug Development In January 2026, AstraZeneca acquired Modella AI , following a collaboration announced in July 2025. Modella’s pathology foundation models and AI agents support oncology research, biomarker discovery and clinical development. AstraZeneca has also identified partnerships with Tempus and Pathos as part of its work on AI models that combine images, molecular data and clinical text. Earlier discovery agreements include a December 2023 collaboration with Absci to develop an AI-designed antibody

## Interactive Brokers Built a Way for AI Agents to Trade Without You

Date: 2026-10-22T05:41:25.173Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/interactive-brokers-built-a-way-for-ai-agents-to-trade-without-you

Interactive Brokers used its Q3 2026 earnings call on October 15, 2026 to disclose that clients can now hand trading decisions to an AI agent. "We also launched autonomous AI trading, which allows eligible clients to connect the AI agent of their choice," CEO Milan Galik said. Clients create and fund a dedicated partition within an existing account in a single session, without opening a new account, according to the company. The feature works with any compatible MCP client or AI platform and supports trading in global stocks and options across day, overnight, and overnight-plus-day sessions. It is available in several countries, with additional countries to follow. The rollout is limited for now. Initial invitations went to clients who had already used the firm's human-in-the-loop AI connector, and Galik said about 3,000 clients were invited over the preceding several days, with eligibility expanding in the following weeks. Galik described the AI connector that preceded autonomous trading as human-in-the-loop, and said it "remains 1 of the fastest adopted features we have ever introduced." This quarter, the firm added Perplexity as an official partner and expanded connectivity "to support any AI tool that uses the model context protocol or MCP," including coding and desktop assistants. New connector capabilities include account-level P&L, alert management, watch lists, and trading of two-leg option strategies. Wells Fargo analyst Kyle Voigt asked what share of the customer base would be eligible and whether AI tools would increase trading activity. Galik did not give a percentage, but said the invited number "is going to greatly increase." He said he expects clients to grow more comfortable with AI making trading decisions on their behalf, adding, "we expect higher trading volumes related to AI adoption over time." Internally, the firm is rolling out AI-assisted tools to its client service team that "retrieve data across multiple internal systems" and analyze client requests for representatives. "New account onboarding, compliance, and surveillance are all incorporating AI and machine learning to help us scale," Galik said. CFO Paul Brody described the firm's core strategy as "automating as much of the brokerage business as possible." Commission revenue reached $602 million, up 12% year over year, and GAAP diluted EPS was $0.69, up 17%.

## OpenAI Wants to “Accelerate” the Work Behind New Drugs

Date: 2026-10-22T05:02:00.654Z
Author: Amit Raja Naik
Desk: Pharma
URL: https://aimmediahouse.com/pharma/openai-wants-to-accelerate-the-work-behind-new-drugs

Drugmakers increasingly rely on specialist research firms, manufacturers and internal teams to advance a medicine, while the plans connecting them remain scattered across systems. A delayed study or failed manufacturing batch can unsettle timelines and spending elsewhere. OpenAI’s recent acquisition of Orchestra Bio , a San Francisco-based biotech startup, now takes it into software built to manage those dependencies, giving pharma companies another supplier to assess for the operational work surrounding R&D. OpenAI, in its blog post , said that the deal brings Orchestra’s team and technology into Rosalind, OpenAI’s life sciences models and tools. Financial terms were not disclosed. “Together, we aim to build agents that help teams connect scientific findings with the studies, decisions, and next steps needed to advance a potential medicine,” OpenAI said. OpenAI president and co-founder Greg Brockman also highlighted the application’s potential, writing on X : “Towards acceleration of drug development, an extremely exciting and promising emerging application of our models.” The coordination problem predates the acquisition. Orchestra investors Jay Rughani and Jorge Conde described drug development becoming more distributed across specialist partners. “Yet the work of coordinating those partners and internal departments still largely ran through email, spreadsheets, and manual handoffs,” they wrote in a recent LinkedIn post. Their firm, Andreessen Horowitz, backed Orchestra’s seed financing and is also an OpenAI investor. For pharma’s software suppliers, the deal introduces a model developer with its own specialist application into that coordination market. For drugmakers, the question is whether the combination can improve development decisions, and how it will fit with systems they already use. What Orchestra Brings Orchestra helps teams connect scientific plans, financial forecasts and development milestones. Its agents monitor studies, identify emerging issues and help teams compare responses. OpenAI illustrates the role with a manufacturing setback that delays a clinical trial. The agents identify affected activities and timelines, collect relevant information and compare ways to respond. Scientists and program leaders decide what to do next. That distinction matters to a buyer. The value lies in understanding what a change means across a program, while preserving responsibility for spending, scientific and operational decisions. Founder and CEO Ashoka Rajendra described Orchestra’s work in a 2025 BiotechTV interview as helping teams “manage the intersection of science, finance and operations.” He explained that separate financial and scientific plans can fail to add up to realistic timelines and costs. The software helps teams examine alternatives when experiments,

## Every Major Bank Has an AI Story. None Will Give You the Bill.

Date: 2026-10-21T14:06:26.440Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/every-major-bank-has-an-ai-story-none-will-give-you-the-bill

On the morning of October 13, 2026, journalists and analysts listening to JPMorgan Chase's Q3 earnings call heard CFO Jeremy Barnum say something that would have been unusual at any other point in the AI cycle. The bank's AI spending has been misreported and the actual figure is significantly lower than the $700 million that has circulated in media coverage — and no, he would not say what the real number is. "I'm very reluctant to spend a lot of energy quantifying" AI benefits, Barnum said. He acknowledged that current gains are not yet visible in quarterly earnings. He pointed to early productivity evidence in software engineering, including a test involving legacy code rewrites. He noted that some productivity gains are being offset by token costs. And then he moved on. The same day, Goldman Sachs CEO David Solomon spoke at length about AI on his firm's earnings call. He cited studies estimating the U.S. data center buildout could require more than $10 trillion of investment through 2032. He described Goldman's position in the AI cycle as a financier of the infrastructure those systems require, not a developer. He laid out a framework for responsible AI governance. He did not say a single thing about what Goldman is spending on AI internally, what it is getting for that spend, or whether any of it is showing up in results. The following day, Bank of America CEO Brian Moynihan said the bank has approximately 130 AI projects in production, each with a defined payback, and that AI-related benefits are running at $250–300 million this year. Moynihan said benefits are expected to be two to three times the expense involved over time. He did not say what the expense involved currently is. What Each Bank Said, and Didn't The Q3 2026 earnings calls produced a consistent pattern across all five institutions. Banks were willing to discuss AI in terms of operational outcomes — faster code, more efficient advisers, fewer manual processes. They were less willing to discuss what they are paying for those outcomes. And they were almost entirely unwilling to give analysts the pair of numbers — cost and benefit — that would allow an outside observer to evaluate whether the returns are real. Citigroup CFO Gonzalo Luchetti said the bank deploys approximately $2 billion per year in technology investment, with AI layered on top. The "layered on top" construction was

## A $284 Billion Bank Just Built an AI Operating System for Small Businesses

Date: 2026-10-21T10:42:11.060Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/a-284-billion-bank-just-built-an-ai-operating-system-for-small-businesses

The Huntington National Bank announced The 1866 Studio on October 19, 2026, a new AI-powered operating system for aspiring and existing small business owners built with Gemini platform, Google Cloud's AI platform. Named for the year Huntington was founded, the platform brings business guidance, financial intelligence, digital tools, and human expertise into a single, connected experience. The 1866 Studio provides agentic AI workflows across the financial, operational, and management needs of business owners, according to the company. By retaining relevant business context across tasks, it is designed to let owners orchestrate decisions across disparate tools while maintaining control as their needs evolve. Brant Standridge, president of Huntington Bank, said helping businesses thrive has been central to Huntington's purpose for more than 160 years. He described the platform as "bringing together Huntington's expertise and Gemini platform to help owners make more informed decisions and spend more time building the businesses they imagined." The initial capabilities include three agentic workflows. Business launch support uses built-in AI agents to guide owners through entity selection and registration, licensing, and operational setup, and generates a customized action plan to move from idea to launch. Marketing and campaign creation uses multimodal Gemini models to generate tailored marketing collateral, digital ad copy, and social media campaigns. Loan readiness assistance uses AI-powered agents to evaluate financing readiness by identifying potential gaps, explaining lender expectations, and recommending actions to strengthen eligibility for future financing. "Small business owners shouldn't have to become experts in every function or in AI to build a successful business," said Vijay Konduru, head of innovation and AI products at Huntington. He said the platform is intended to increasingly support "the full range of those jobs across the business lifecycle, from turning an idea into a business to running and growing it, and ultimately preparing for succession, sale or exit." Konduru said the agentic workflows sit alongside personalized expert content and an AI assistant that helps owners monitor business operations and finances and manage daily work, "while they remain in control." The platform is built on Google Cloud's platform to build, scale, govern, and optimize agents, combined with Google's multimodal Gemini models. It extends Huntington's multi-year strategic partnership with Google Cloud to modernize technology , advance governed AI capabilities, and accelerate innovation. "By using Gemini platform and Gemini frontier models to build The 1866 Studio, Huntington is creating an innovative experience that helps small business owners

## A Brooklyn Restaurant Owner Built Her Own AI Finance Stack

Date: 2026-10-21T08:48:07.663Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/a-brooklyn-restaurant-owner-built-her-own-ai-finance-stack

Angela Moznerova had one analysis on her list for years that never got done: a full review of her restaurant's multi-level recipe bill of materials and plate-level cost of goods sold. Moznerova works as a fractional CFO and also owns a restaurant in Brooklyn. "Like most owners, I let the urgent beat the important," she wrote in a LinkedIn post on October 20, 2026. She built it during the AI Finance Accelerator run by AI Finance Club, using Python with AI acting as what she called her "junior analyst." The result covers more than 100 ingredients and sub-recipes, costed down to the plate and refreshed every week. Moznerova then extended the system beyond costing. She added a Claude skill that pulls sales from platform websites into the restaurant's daily sales flash, checks every figure, and logs the weather, "so a rainy Friday isn't mistaken for a trend." She then built a sales forecast for each channel using Python and AI, tested it against the restaurant's sales history, and linked it to plate costs. The three pieces now run as one system. "Costing, daily sales and forecasting now form one picture of the business," she wrote. On cost, Moznerova said the approach changed what was feasible for a business of her restaurant's size. "Python with AI made this affordable for a business our size," she wrote. "Analysis I could never justify paying for took days to build." Moznerova built verification into every step. "Automation is only as good as its controls," she wrote. "Every run checks itself against source data." That check surfaced a specific error. The model flagged a recipe that implied 21 lb of pepper, while invoices showed the restaurant had bought 5 lb. Moznerova cited the case under a broader point about how faster tooling affects decision-making: "Speed raises the stakes on judgment." She credited members of the accelerator's fifth cohort with shaping how she built the system. Tiffany Stewart, CPA, CMA, showed her how to organize her Claude and ChatGPT setup. Questions from Jan Hermans and insights from Marianne Neuper, CPA, CMA, reminded her that "data cleaning needs financial judgment." Her coach, Bastian Kneuse, asked whether another restaurant could use the system. "I think the answer is yes," Moznerova wrote. "Small businesses and startups have always had the same questions as large companies. With the right finance partner and these tools, the answers are within their reach."

## Wells Fargo Hires BBVA's Enterprise AI Lead to Run AI and Data for Consumer Banking.

Date: 2026-10-21T06:28:22.545Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/wells-fargo-hires-bbvas-enterprise-ai-lead-to-run-ai-and-data-for-consumer-banking

Wells Fargo announced on October 15, 2026 that Cayetano Gea-Carrasco will join the company as head of AI and Data for Consumer Banking and Lending. He will report to Saul Van Beurden, who serves as both head of Enterprise AI and co-CEO of Consumer Banking and Lending, and will be based in San Francisco. The role covers the full lifecycle of AI use cases within the division. Gea-Carrasco will be responsible for simplifying the data environment and accelerating the responsible delivery of AI for Consumer Banking and Lending. He will also work with Wells Fargo's Enterprise AI team to scale solutions across the wider company. "Customers do not think about channels. They expect us to know them, make banking easier, and be there when they need us," Van Beurden said. "Cayetano understands how strong data and disciplined execution turn AI into better customer experiences. His track record will help us move faster while keeping safety and sound risk management front and center." Gea-Carrasco joins from BBVA, where he led the bank's enterprise AI transformation and its global data and analytics organization. Before BBVA, he spent 15 years at Moody's in senior roles spanning software, data, and analytics. "Wells Fargo's scale and depth of customer relationships create a powerful opportunity to put AI to work," Gea-Carrasco said. "I'm joining to make everyday banking simpler for our customers and deliver measurable business outcomes, with the controls and accountability our customers expect." The appointment follows Wells Fargo's Q3 2026 earnings call on October 13, where CEO Charlie Scharf said the bank had risen to #5 globally and #3 in the U.S. on the Evident AI Index. Scharf described AI deployments across code writing, branches, marketing campaigns, and cybersecurity, and said the bank had launched an AI-powered tool for its financial advisers. Wells Fargo has approximately $2.3 trillion in assets and operates through four reportable segments: Consumer Banking and Lending, Commercial Banking, Corporate and Investment Banking, and Wealth and Investment Management.

## Capital One Rebuilt Its Tech Stack for AI

Date: 2026-10-21T05:52:18.650Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/capital-one-rebuilt-its-tech-stack-for-ai

During Capital One's Q3 2026 earnings call on October 20, 2026, CEO Richard Fairbank described its multi-year technology transformation as an AI project from the start. "We set out to rebuild our technology from the bottom of the tech stack up," Fairbank said, adding that "the objective function of the rebuild was AI." The rebuild began in 2013, and Fairbank said the bank was working backwards from "having a modern tech company powered by AI" in "a world where everything was going to be about AI in real time." He acknowledged the bank did not anticipate how the technology would develop. "We did not predict generative AI, and we did not predict agentic AI, yet AI was the objective function of this transformation," he said. Had the bank known, he said, "we would have built exactly what we built." Wells Fargo analyst Don Fandetti noted that Capital One "ranked pretty well on an AI industry report" and asked whether the bank's AI investments were producing specific revenue or market share gains, and whether management would put numbers on them. Fairbank declined. "We certainly do not intend to quantify the benefits," he said. He did describe where the bank expects AI to have an effect. "AI will be transformative for the customer experience," Fairbank said, pointing to "the opportunity with modern technology data and AI to transform the entire data ecosystem" inside banks. "The pace of advancement in AI is just breathtaking," he said, adding that "companies will adopt AI and get a lot of benefits." In response to Piper Sandler analyst Bill Carcache on how the franchise is positioned for a future credit downturn, Fairbank said the bank uses "AI-based monitoring that is full file" in some areas. "Instead of manual, it is automated," he said, while noting it is still "the early days" for some of that monitoring. Fairbank said in prepared remarks that Capital One continues to "lean into investments in modern technology, data, and AI." Responding to Bank of America analyst Mihir Bhatia, he said much of that investment is already reflected in reported results and will continue to increase, including further spending on the Discover network. As of the end of September, 100% of new Discover card originations ran on Capital One's tech stack, with the back-book conversion expected to finish in the first quarter, according to the company. Capital One reported Q3 2026 net income

## 59% of Enterprises Say AI Waste Rose. Only 31% Know What They Spend.

Date: 2026-10-21T04:34:36.071Z
Author: Sachin Mohan
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/59-of-enterprises-say-ai-waste-rose-only-31-know-what-they-spend

Enterprise AI budgets are still going up. The ability to see where that money goes is going down. Flexera surveyed 512 technology professionals worldwide for its 2026 State of ITAM Report . The main finding was about waste. 59% of organizations said wasted AI software spend increased over the past year. Only 7% said it decreased. No other category came close. Public cloud software, the category that dominated waste conversations for most of the last decade , came in at 44%. SaaS came in at 43%. IaaS and PaaS came in at 41%. AI is now the fastest-growing source of wasted technology spend in the enterprise, and it got there in the same year most organizations started taking it seriously. The visibility numbers explain why. Only 31% of organizations have accurate visibility into their AI software. Complete visibility across the entire IT estate fell to 36%, down from 43% in 2025. Only 29% measure the value of the AI software they are paying for. "AI is changing the economics of IT faster than most organizations can adapt," said Becky Trevino, chief product officer at Flexera. The Cost Problem Has Already Changed Decisions KPMG's Q2 2026 Global AI Pulse , which surveyed 2,145 senior leaders across 20 countries at organizations with more than $50 million in annual revenue, shows what happens when the cost picture stays unclear. 49% of organizations said they have delayed, paused, or scaled back AI agent deployments because expected costs outweighed anticipated benefits. Roughly a quarter narrowed deployments. A similar share delayed further rollout. These are not organizations abandoning AI. They are organizations that reached production and found the running costs were not what the pilot suggested. Only 35% of organizations in the KPMG survey report full visibility into their AI operating costs. The rest are working from partial, delayed, or fragmented information. The difference that visibility makes is measurable. Organizations with full cost visibility are five times more likely to report established ROI than those without it, 15% compared with 3%. The controls that would produce that visibility are not yet standard. Slightly more than half of organizations include cost reviews in AI approval processes. 40% use token or usage budgets. Fewer than 40% have architecture or prompt-design standards for efficiency. Flexera's data shows the same gap from the asset management side. 84% of respondents named tracking and adopting AI applications as a top challenge, the

## Chipmakers Turn to Upgrades as AI Demand Outpaces Expansion

Date: 2026-10-19T07:51:52.274Z
Author: Amit Raja Naik
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/chipmakers-turn-to-upgrades-as-ai-demand-outpaces-expansion

ASML expects its planned equipment output to fall short of customer demand in 2027 and 2028. Upgrades to machines already installed in chip factories are becoming a growing source of business. At its latest earnings call , the Dutch equipment maker said strong demand for productivity upgrades helped lift third-quarter service and field-option sales to approximately €3.3 billion, almost €400 million above guidance. The upgrades cover both extreme ultraviolet, or EUV, and deep ultraviolet, or DUV, lithography equipment. “The ability to enable an increase in capacity through upgrades on the same footprint is very compelling for our customers, as the effect is an immediate increase in output capability,” said Roger Dassen , chief financial officer at ASML. US chipmakers are pursuing factory improvements alongside expansion. In Intel’s second-quarter results , CFO Dave Zinsner attributed stronger volumes partly to “higher factory yields and improved cycle times.” Intel is also “meaningfully increasing our investments in equipment, clean room space, and substrates,” he said. Micron is modernizing existing facilities while building new fabs. Its more than $2 billion Manassas expansion introduced 1-alpha DRAM production in May, a transition the company said would quadruple its DDR4 wafer supply at the Virginia site. Separately, Micron said advanced EUV tools would help it “optimize both cleanroom space efficiency and patterning” as it develops future DRAM technology. Its annual report projects first wafer output from its new Idaho fabs in mid-2027 and late 2028. Demand is Running Ahead of New Machines “Supply shortages across advanced logic and memory nodes continue to increase in size and in time horizon,” said Christophe Fouquet , chief executive officer at ASML. ASML expects memory net system sales to grow around 75% this year, driven by customer expansion and increased EUV use in advanced DRAM production. Advanced logic net system sales are expected to grow more than 25%. The company is increasing its own output. It expects to produce roughly 65 low-NA EUV systems in 2026, compared with 44 shipped last year. It plans a 30% capacity increase in 2027 for both low-NA EUV and DUV immersion systems, and is investigating an increase beyond 30% in 2028. Customers are expressing interest in orders for 2029 and 2030, although ASML is still assessing demand and capacity before accepting those commitments. Dassen said many installed tools had yet to receive upgrades, leaving further opportunities to increase production from existing equipment. “We will continue to

## Bank of America’s 130 AI Projects Are Delivering Up to $300 Million This Year

Date: 2026-10-16T10:12:35.430Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/bank-of-americas-130-ai-projects-are-delivering-up-to-300-million-this-year

Bank of America CEO Brian Moynihan used the Q3 2026 earnings call on October 14, 2026 to put specific numbers on what AI is delivering inside the bank, and what it expects to deliver next year. The bank currently has approximately 130 AI projects in production, each with a defined payback. AI-related benefits are running at 250–300 million this year and are expected to reach $700 million in 2027, based on the pipeline of projects already in place. Moynihan said benefits are expected to be two to three times the expense involved over time. The 10% efficiency gain Moynihan cited is tied directly to the bank's technology development spend, which runs at approximately $4 billion per year. "We are 10% more efficient," he said, "so next year we get 10% more work for the same cost." CFO Alastair Borthwick also noted that headcount has fallen as the company automates processes and simplifies work, describing technology modernization and AI as explicit drivers of that reduction alongside investment in client acquisition and franchise growth. Erica, Global Banking, and Wealth Management Moynihan described Erica, the bank's widely used virtual assistant, as now capable of operating on an agentic basis, helping consumers complete tasks that previously required a human intermediary. He said the bank is working with agent providers to ensure that customers remain in control, and that human approval is required before any money movement takes place. "We believe strongly in ‘human in the loop’," he said. Borthwick added that "Erica, Zelle, and AI-enabled tools are helping clients and helping employees" complete activity faster and more consistently, describing the tools as operating across both customer-facing and internal workflows in consumer banking. Revenue came in at $11.6 billion, up approximately 4% year over year. In Global Banking, Borthwick disclosed that the bank launched a GenAI tool in September designed to make prospecting and onboarding more efficient for commercial bankers, framing it as a direct productivity lever for the client acquisition process. In Wealth Management, where revenue grew approximately 13% to $7.1 billion, Borthwick said AI tools "continue to expand our capacity to deliver personalized advice at scale," crediting advisor productivity and digital engagement alongside AI for driving the segment's growth. The Broader Picture Bank of America reported Q3 2026 net income of $6.9 billion and EPS of $0.91 on revenue of approximately $30 billion, up around 3% year over year. The results include a

## Wells Fargo’s CEO Thinks AI Agents Will Reshape Banking

Date: 2026-10-16T07:00:38.517Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/wells-fargos-ceo-thinks-ai-agents-will-reshape-banking

Wells Fargo CEO Charlie Scharf used the bank's Q3 2026 earnings call on October 14, 2026 to lay out how the firm is deploying AI internally and where it sees the clearest near-term opportunities — while also addressing analyst questions about whether AI agents could become a meaningful risk to deposit stability. Scharf said Wells Fargo's AI focus was recently recognized in the Evident AI Index, which ranks banks across four pillars: talent, innovation, leadership, and transparency. The bank ranked #5 globally and #3 in the United States. On internal deployment, Scharf described using AI to write code faster, to support branch and marketing activity, and to identify and close cybersecurity gaps — specifically, "to help identify as well as write and test code to close previously unknown cyber vulnerabilities." Externally, he said the firm is using AI to improve the customer experience and is evaluating how "third-party models and agents can provide value-added capabilities to our clients." They also introduced a new AI-powered tool for financial advisers in the Wealth and Investment Management business, built on BlackRock's Aladdin Wealth technology. Scharf described it as designed to streamline routine work and said it is the first of several planned capabilities under the bank's broader WIM AI strategy. Wells Fargo's WIM business had approximately $2.5 trillion in client assets at quarter end. On the question of AI agents and deposits — raised by an analyst who asked whether agents could accelerate cash movement to higher-yield accounts — Scharf pushed back. He said high-yield alternatives have existed for years and are already easy to access, that consumers are unlikely to give agents unilateral control over their money in the near term, and that customers keep deposits at Wells Fargo for reasons beyond yield. "There's no question that agents will become a part of the way people interact," he said, but added that a mass movement driven by agents "is completely overdone" as a near-term concern. He said the bank is building its own agents to serve customers better and retain that relationship. Scharf also addressed AI's implications for cost structure directly. "This idea of scale matters in an AI world is even more so," he said, arguing that a bank of Wells Fargo's size can spread the cost of building an agent across a far larger operation than a smaller competitor could. CFO Michael Santomassimo listed AI-related disruption among the external risks

## Albertsons CEO Says the Company’s Four AI Bets Are Already Delivering.

Date: 2026-10-15T10:13:44.320Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/albertsons-ceo-says-the-companys-four-ai-bets-are-already-delivering

Albertsons CEO Susan Morris used the company's Q2 2026 earnings call on October 13, 2026 to lay out where each of the grocer's four AI initiatives stands — and to push back against any suggestion that the company is waiting on infrastructure before it moves. "AI is absolutely essential, complementary, and we're not waiting," Morris said. "Some of it will come later, but some of it's helping us right now." Morris described the four bets in direct response to an analyst question about whether Albertsons has the right technology in place to execute its current priorities. Each initiative, she said, is specifically designed to support improvements in store standards. The operational baseline the company is focused on as it works through a challenging sales environment. Merchandising intelligence is aimed at enabling faster, smarter decision-making on category management, covering assortment, pricing, and promotions. Store labor productivity is focused on driving efficiency in one of the business's most significant cost lines. Demand forecasting, which Morris described as running "end to end, from farm to table," is designed to improve in-stock results by strengthening the company's planning capabilities upstream. The fourth initiative, inventory optimization, was referenced in the context of demand planning but not separately detailed on the call. Executive Chair Meg Whitman added a top-line frame: "We've built a strong e-commerce platform and see AI as critical to our success." No AI-specific financial results, cost savings, or investment figures were disclosed on the call. The broader financial picture for the quarter was mixed. Net sales of $18.89 billion were down 0.1% year over year, while digital sales grew 11% and digital penetration reached 10.5%. Adjusted EBITDA came in at $784 million. Management withdrew its full-year outlook and said a multiyear plan would be presented by the fourth-quarter call.

## Citi Is Deploying AI at Scale. Jane Fraser Thinks It Deserves More Credit for It.

Date: 2026-10-15T09:44:46.978Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/citi-is-deploying-ai-at-scale-jane-fraser-thinks-it-deserves-more-credit-for-it

Citigroup reported its best third-quarter revenue in a decade on October 13, 2026, with net income up approximately 40% year over year. But the most substantive exchange on the earnings call was how far Citi has actually gone in deploying AI, and what it has to show for it. CEO Jane Fraser used her prepared remarks to lay out how Citi thinks about AI across four dimensions — growth, efficiency, employee training, and cyber defense. She described the bank's multi-year technology transformation as having delivered the foundation needed to "implement AI firm wide," and said the firm is now seeing that investment translate into measurable operational outcomes. CFO Gonzalo Luchetti said Citi is deploying approximately $2 billion per year in technology investments, directed at both maintaining core capabilities and building on them, with AI layered on top driving further process improvements. What AI Has Changed Inside Citi The firm's primary agentic coding tool, Devon, is now used by 14,000 engineers across the bank. Fraser said this adoption contributed to a 30% year-over-year reduction in tech vendor spend and accelerated new product launches by 30–40%. Beyond engineering, Fraser said the bank has identified around 100 of its most manual processes and has either completed or is in the process of re-engineering them end to end using AI and automation, framing it as a systematic effort to remove structural inefficiencies rather than apply AI at the margins. On the client-facing side, AI tools for client service agents are now live in two countries, with response times improving by 21%, according to the company. Luchetti, on the expense question, described AI and automation as the mechanisms through which the firm expects to unlock structural efficiencies over time, the kind that show up in the expense base rather than in any single quarter. CFO Gonzalo Luchetti said Citi is deploying approximately $2 billion per year in technology investments, directed at both maintaining core capabilities and building on them. Fraser added that AI is layered on top of that base, driving further process improvements. Luchetti, in response to an analyst question on expenses, described AI and automation as the mechanisms through which the firm expects to unlock structural efficiencies over time. On the Broader Ecosystem Fraser said Citi has not seen a slowdown in investment in the AI ecosystem. She flagged compute and energy as the primary bottlenecks and said continued investment to address those

## Goldman Sachs Sees the AI Infrastructure Buildout as a $10 Trillion Financing Opportunity

Date: 2026-10-15T09:02:47.676Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/goldman-sachs-sees-the-ai-infrastructure-buildout-as-a-10-trillion-financing-opportunity

Goldman Sachs reported Q3 2026 net revenues of $18.7 billion on October 13, 2026 — up 23% year over year and the second-highest quarterly figure in the firm's history. Diluted EPS came in at $17.01, up 39%, with return on equity at 18.7% and return on tangible equity at 20.5%. Against that backdrop, CEO David Solomon used the call to lay out Goldman's view of the AI infrastructure investment cycle as one of the firm's most significant near-term opportunities. Solomon cited external studies estimating that the U.S. data center buildout could require more than $10 trillion of investment through 2032. At that scale, he said, the annual capital requirement would represent approximately 3.6% of U.S. GDP. Demand for compute currently exceeds supply, he said, and that imbalance is what creates financing opportunities for Goldman across its capital markets, lending, and advisory businesses. In regards to where Goldman sits in the AI investment cycle, Solomon said they are not a builder of AI systems, but a financier of the infrastructure those systems run on. Total financing revenues reached a record $12.7 billion in the first nine months of 2026, representing over 40% of combined FICC and equities revenues. Equities financing revenues in Q3 alone hit a record $3.37 billion, up 96% year over year, driven by elevated prime balances and Asia activity. At the same time, Solomon acknowledged the limits of visibility into how the cycle evolves. "No one knows how the demands for capital and compute will evolve as the broader AI ecosystem matures," he said. He also noted that rising rates could pressure project returns and that the cycle will at some point "recalibrate." Solomon also addressed AI governance directly, framing it as a subject where Goldman's existing risk management practices are directly applicable. He named rigorous testing, independent review, continuous monitoring, and clear human accountability as the framework for responsible deployment. He acknowledged "the accelerating pace of AI technology and the unchecked advancement of frontier models" as a broader concern, but argued the answer is not to halt progress. Analysts pressed on multiple AI-related angles during the call. Evercore ISI's Glenn Schorr raised the question of how agentic trading, tokenization, and prediction markets could affect Goldman's markets business over time. Bank of America's Ebrahim Poonawala asked whether financing has become the real constraint on AI capital expenditure. Solomon said the firm is watching both dynamics closely.

## JPMorgan Expands AI Agent Use While Blocking External Access

Date: 2026-10-14T11:05:23.421Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/jpmorgan-expands-ai-agent-use-while-blocking-external-access

During JPMorgan Chase's Q3 2026 earnings call on October 13, 2026, CEO Jamie Dimon confirmed JPMorgan deploys AI agents internally and intends to expand that use for customers — while naming rogue agents as a distinct operational concern that the firm is actively building defenses against. "We are not gonna allow just any outside agent to come inside the company," Dimon said. He described the risk as significant given the potential scale. "You do not want rogue agents, and there could be thousands of them," Dimon said. He added that regulators should be considering whether the proliferation of bots and agents could create systemic harm. CFO Jeremy Barnum pushed back on figures that have circulated in media coverage of JPMorgan's AI spending. He said the actual AI expense number is "significantly less" than what has been reported, referencing a figure of approximately $700 million annually. He also noted the earlier benefit quantification that accompanied those expense figures was accurate and included established machine learning applications such as fraud prevention. Barnum said he is "very reluctant to spend a lot of energy quantifying" AI benefits and that current gains are not yet visible in quarterly earnings. He pointed to early productivity evidence in software engineering, including a test involving legacy code rewrites, and noted that some productivity gains are being offset by token costs. On infrastructure strategy, Barnum described an approach built around avoiding lock-in: the firm uses both proprietary and open-weight models and applies routing logic to direct queries to the most cost-effective suitable model. "Model agnosticism, harness agnosticism, and ability to run both proprietary models" was how he framed the approach. Bank of America analyst Ebrahim Poonawala raised the question of whether AI-powered personal agents could accelerate cash movement out of JPMorgan accounts. Barnum framed the risk as the intersection of two existing dynamics, yield-seeking behavior and fintech competition, rather than a structurally new threat. He said the firm is not complacent and noted that a "smart cash" product went live in recent weeks as a pilot response. On the question of customer protection, Barnum said customers who share credentials with an external agent bear the risk if that agent acts against their interests. Barnum also addressed AI concentration risk in JPMorgan’s lending book, describing hyperscaler-related exposure as “quite modest” relative to total wholesale exposure. He noted the AI theme has diffused broadly enough to reach working capital revolvers

## "Technological Progress Outpaced Adoption" — The Real Lesson From a Decade of AI in Pharma

Date: 2026-10-14T06:10:20.773Z
Author: Sachin Mohan
Desk: Pharma
URL: https://aimmediahouse.com/pharma/technological-progress-outpaced-adoption-the-real-lesson-from-a-decade-of-ai-in-pharma

Alessandro De Luca has been watching AI move through Merck Group for nearly a decade. As Head of Digital Enterprise Solutions and Group CIO, he was talking about self-driving supply chains in 2016, when most pharmaceutical companies were still debating whether to invest in basic automation. Ten years later, he has a clear-eyed view of what actually proved hard. "Looking back I would say that convincing humans to trust autonomous supply-chain decisions was harder than teaching machines to decide," De Luca told AIM Media House. "The biggest obstacles were explainability, accountability, governance, risk management, and organizational change." The machines, he says, improved quickly. The people took longer. The Governance Gap Behind Pilot Failures The observation speaks directly to one of the most documented problems in enterprise AI: the gap between what AI systems can do in a controlled pilot and what they actually deliver when embedded in daily operations. "AI has fallen short when initiatives were driven by hype rather than clear business need. Poor data quality, weak process readiness, limited governance, and low user adoption can prevent even accurate models from creating value," De Luca said. Merck's response was to treat AI not as a technology project but as a business transformation agenda. That meant assembling cross-functional teams of business leaders, IT, data scientists, compliance staff, and domain experts around outcomes that could actually be measured. "The biggest shift was cultural," De Luca says. "Scaling AI depends on trust, business ownership, and accountability for outcomes." Where Merck Is Actually Seeing Returns On the question of where AI has delivered real value, De Luca is specific. Demand forecasting, inventory optimization, predictive maintenance, quality control, and clinical documentation are the areas he names as producing measurable gains in productivity, cycle time, and service levels. They are problems with well-defined parameters, strong underlying data, and outcomes that can be tracked directly against business metrics. The company's SMARTfacturing platform sits at the center of this effort. Built around AI-driven scenario simulation, predictive analytics, real-time monitoring, and decision support, the platform is designed to make manufacturing operations more responsive and supply chains more resilient. The Bengaluru capability center plays a central role in building and scaling these systems across Merck's Life Science, Healthcare, and Electronics business sectors. On how Merck measures returns, De Luca draws a clear line between use case types. Manufacturing and supply chain ROI is tracked through cycle time, yield, forecast

## Blend's AI Mortgage Agent Is Now Processing 3,500 Loans a Week

Date: 2026-10-13T11:18:51.999Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/blends-ai-mortgage-agent-is-now-processing-3500-loans-a-week

Blend Labs announced on October 12, 2026 that a tenth lender has joined Blend Autopilot, its agentic AI platform for home lending. The milestone comes six months after Autopilot began processing live loans and three months after it became commercially available. The platform is currently working on more than 3,500 new loans per week, an annualized pace of nearly 200,000 loans, according to the press release. Autopilot was built around a specific structural problem in mortgage origination: more than half of borrower activity happens outside of standard business hours, leaving document submissions sitting unreviewed for hours or longer. The platform addresses this by reviewing documents as they are uploaded, comparing them against lending guidelines, identifying missing or inconsistent information, and generating follow-up requests while borrowers are still engaged with their application. The agents handle work across document review, income review, borrower follow-ups, pre-underwriting, and closing quality control, according to the press release. Previously reported production results show 10% to 15% higher pull-through rates, cycle times shortened by two to four days, and 4.5 hours of fulfillment work automated per loan on average. Onity Mortgage, which serves 1.4 million customers and is one of Autopilot's earliest adopters, described the operational impact specifically. "Customers are uploading documents when it's convenient to them — this is weekends, this is after our business hours — and when they upload a document, instead of the document sitting there and being picked up hours later, Autopilot's pre-underwriting agent is reviewing that automatically," said Patrick Joyce, VP of Digital Products at Onity Mortgage. Onity is also expanding its use of Blend's MCP Server to power voice agents that can answer routine application questions around the clock. The latest lenders to sign on include Whatcom Educational Credit Union, one of Washington's largest credit unions, and Gate City Bank, alongside other community financial institutions and regional banks. Lenders can deploy Blend's purpose-built agents or build their own through Blend's MCP Server, which provides access to the company's lending data, 150-plus integrations, and workflow controls built over nearly 15 years. In 2025, Blend helped financial institutions process nearly $1.3 trillion in loan applications, according to the press release. "We're moving fast, our customers are moving fast with us, and together we're leading the agentic transformation of lending," said Nima Ghamsari, Co-founder and Head of Blend.

## Meta Unveils Muse Wonder, Its Most Advanced Agentic AI Model

Date: 2026-10-13T11:06:18.884Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/meta-unveils-muse-wonder-its-most-advanced-agentic-ai-model

Meta introduced two new AI models on October 12, 2026 — Muse Wonder and Muse Wonder Full — built for complex personal and professional tasks including software development, legal document interpretation, insurance claim processing, and healthcare information. The models were developed by Meta Superintelligence Labs and trained at Meta's Prometheus data center in Ohio, according to the press release. Both models build on the earlier Muse Spark, launched this year, and represent Meta's bid to compete directly with frontier models from OpenAI, Anthropic, and Google. Meta CEO Mark Zuckerberg told the Financial Times that the company has reached the AI frontier and is "already working on the next step up." Muse Wonder introduces two capabilities central to its agentic design. Active Memory allows the model to retain and retrieve relevant context across conversations, enabling it to track user preferences, prior instructions, and ongoing tasks. Active Focus keeps the model working through obstacles: trying alternative approaches, recovering from errors, and checking in with users before taking consequential actions rather than stalling. "When you do real-world tasks, you're going to get roadblocks and you're going to get things that come up or various setbacks," said Alexandr Wang, Meta's AI chief. "We specifically tried to train the model to be able to try other approaches instead of stalling and then checking back in with the user at any time." Meta said Muse Wonder outperformed competing models including Fable, Opus, and Astra on MemoryBench, which tests memory construction and continual learning. Muse Wonder is available to developers through the Meta Model API and to Muse app users through paid subscription tiers. Muse Wonder Full carries no public API and is restricted to paid Muse Code subscribers. Wang said API pricing for Muse Wonder will remain "meaningfully below" comparable offerings from OpenAI and Anthropic. On safety, Meta said Muse Wonder offers its strongest protections to date against CBRN and cyber misuse, and has been trained to handle private information with discretion. Those safeguards are currently limited to the paid tier. Some users have reported that the Muse agent accessed personal messages without authorization — a claim Meta has disputed, with Wang attributing incidents to users forgetting earlier permissions they had granted.

## AMD Built a Rack Server Around What AI Agents Actually Need to Run

Date: 2026-10-13T09:58:33.950Z
Author: Sachin Mohan
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/amd-built-a-rack-server-around-what-ai-agents-actually-need-to-run

AMD introduced the Helion agentic AI rack on October 12, 2026, positioning the product as purpose-built infrastructure for the distinct compute demands of agentic workloads. The announcement was made at the OCP Global Summit in San Jose. According to the company, the design premise behind Helion is that agentic AI does not map cleanly onto existing infrastructure categories. "Agentic AI is not a single workload. It spans personal agents, enterprise applications, code execution and complex workflows, each with different compute requirements," said Dan McNamara, Senior Vice President and General Manager of Compute and Enterprise AI at AMD. The rack is built to let customers and partners configure compute accordingly rather than applying a fixed hardware formula across different deployment types. Helion is an OCP spec-based rack blueprint available in liquid- and air-cooled versions. Two configurations define its range, according to the press release. The first is a CPU-only rack powered by 6th Gen AMD EPYC processors, supporting up to 32,768 CPU cores, designed for hyperscale customers who want dedicated capacity for agent execution while connecting to separate model inference infrastructure. The second is an all-in-one rack combining AMD EPYC CPU nodes, AMD Instinct MI350P GPU nodes, and general-purpose compute on a single platform, designed for enterprise agentic AI deployments. The platform includes an integrated token router that directs agent model calls across frontier and open-source models . AMD said it plans to contribute the Helion specification to OCP for broader community development. Eight partners have confirmed support for the platform at launch: Cisco, Dell Technologies, Hon Hai Technology Group, HPE, Lenovo, QCT, Supermicro, and Wiwynn. Rack configurations are currently under development and are planned to be available in November 2026. Dell described Helion as extending its PowerRack platform to agentic AI, with configurations that combine AMD's rack-scale architecture with Dell's engineering alongside networking and storage options. HPE framed its involvement as a continuation of its longstanding partnership with AMD on open standards, noting support for both air-cooled and direct liquid-cooled networking. Supermicro said it would bring Helion to market through its Data Center Building Block Solutions. Helion follows AMD's Helios rackscale solution and represents the company's bid to bring the same open-platform approach to the agentic infrastructure market, where Nvidia currently leads with its own rack reference designs including the NVL72.

## Hilton Hires Uber's Jeff Moore as CTO to Accelerate Its AI and Platform Strategy

Date: 2026-10-12T11:48:12.105Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/hilton-hires-ubers-jeff-moore-as-cto-to-accelerate-its-ai-and-platform-strategy

Hilton named Jeff Moore as Executive Vice President and Chief Technology Officer on October 6, 2026, bringing in a senior engineering executive from Uber as the hospitality company moves to accelerate its AI agenda. Moore starts October 26 and will report directly to Chris Nassetta, president and CEO, as a member of Hilton's Executive Committee. The hire closes a search Hilton began in May, when Nassetta flagged AI as a significant opportunity ahead for the company. Hilton has been laying the groundwork. The company launched the Hilton AI Planner, an AI-powered trip planning tool, in beta in March 2026, and has established partnerships with Google, OpenAI, and Anthropic to build out its AI capabilities. Moore joins as those investments move from pilot into broader deployment. Moore brings more than 20 years of technology leadership experience across engineering, product, and operations at global scale. At Uber, he led a global engineering organization of 2,500 engineers supporting Uber Eats, Uber for Merchant, Uber Direct, Uber Advertising, Growth and Marketing, Uber for Business, and Applied AI. Before Uber, he held engineering and product leadership roles at Amazon, including as a founding leader within Fulfillment by Amazon and later overseeing the device portfolio — Kindle, Fire TV, Fire Tablet, and Echo/Alexa products. Earlier in his career he worked in software engineering and operations at Anheuser-Busch. He holds three issued US patents. "At a time of extraordinary change, Jeff Moore's deep technical expertise, operating experience and customer focus will help us move faster, harness AI and next-generation technology," Nassetta said in the announcement. Moore described the AI agenda as a primary draw. "I was especially drawn to Hilton's ambition to harness AI and next-generation technology to reimagine the guest experience and empower team members," he said.

## The RealReal Rolls Out AI Shopping Agent to All 45 Million Members After Pilot

Date: 2026-10-12T11:17:35.323Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/the-realreal-rolls-out-ai-shopping-agent-to-all-45-million-members-after-pilot

The RealReal launched Ask TRR to all 45 million members on October 8, 2026, making its AI shopping agent, built with Google Cloud's Gemini Enterprise, broadly available after a pilot program. This launch is the company's most significant step yet toward replacing keyword search with conversational discovery across its luxury resale marketplace, according to the press release. The case for an AI agent at The RealReal is structurally different from standard retail. The platform lists more than 1.5 million unique items daily, each available as a single unit. Once a piece sells, it is gone. Traditional keyword search works well when a retailer carries multiple units of the same product across time. It works poorly when a shopper knows they want a certain silhouette or era but cannot name the exact designer, and the item they are looking for may sell before the right search term surfaces it. Ask TRR allows shoppers to describe what they are looking for in natural language and receive tailored recommendations built around intent rather than keywords. The agent is built on Gemini Enterprise for Customer Experience and draws on 15 years of proprietary platform data and more than 50 million items previously sold on The RealReal to generate those recommendations, according to the press release. "Ask TRR builds on that advantage, enabling precision and specificity whether you start with a designer name, an occasion, or simply an idea," said Rati Sahi Levesque, president and CEO of The RealReal. Darshan Kantak, VP of Product for Applied AI at Google Cloud, described the direction as AI that goes beyond answering questions toward "understanding people." The RealReal said it plans to extend the capability to product description pages, adding contextual AI recommendations at the point where a shopper is actively considering a purchase, the step closest to conversion. For most retailers, AI-driven discovery is a convenience. For The RealReal, it is closer to a time constraint — every item is a single unit, and inventory turns over daily. A shopper who finds the right piece and waits has no guarantee it will still be there.

## Accenture and Dell Form a Dedicated Business Group to Help Enterprises Run AI on Private Infrastructure

Date: 2026-10-12T09:47:39.252Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/accenture-and-dell-form-a-dedicated-business-group-to-help-enterprises-run-ai-on-private-infrastructure

Accenture and Dell Technologies announced the formation of the Accenture Dell Business Group on October 8, 2026, formalizing a dedicated unit to help enterprises deploy AI on private infrastructure. The move extends a collaboration spanning more than 20 years into a structured offering targeting organizations that need to run AI workloads on their own infrastructure rather than in shared cloud environments. The group targets a constraint that has become more visible as enterprises move AI from pilots to production. For organizations in financial services, life sciences, manufacturing, and government, data sovereignty and compliance requirements limit what can run on public cloud infrastructure. The Accenture Dell Business Group is built around that gap — Dell’s on-premises AI infrastructure paired with Accenture’s implementation expertise. According to the companies. he group's capability set covers four areas. Private AI and neocloud solutions are designed for organizations building AI Factories on dedicated infrastructure with data control and compliance requirements. Integrated private AI stack offerings bring together Dell's full hardware and software portfolio to support agentic AI, inferencing, and high-performance compute workloads. Agentic solutions deployment covers accelerators designed to reduce time-to-value from scaled AI use cases. Open-source and open-weight model support adds task-aware model routing across environments to maintain cost predictability on private infrastructure. Accenture plans to train more than 3,000 practitioners across the Dell portfolio to staff the group. UK-based sovereign AI provider RHAELM is the named early client, with Accenture and Dell having supported the delivery of its sovereign-grade AI solutions for regulated industry clients. "Enterprises will be able to deploy AI solutions faster, align them with their business priorities, maintain greater control through private, hybrid and sovereign environments, and optimize both cost and performance," said Julie Sweet, chair and CEO of Accenture. Michael Dell described the group as helping enterprises "move from AI interest to adoption."

## One in Three Enterprises Deployed an Emergency AI Spending Freeze This Year.

Date: 2026-10-08T09:03:00.000Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/one-in-three-enterprises-deployed-an-emergency-ai-spending-freeze-this-year

Enterprises are not failing at AI because the models do not work. They are failing at AI because the budget process was never built for how AI costs actually behave. Mavvrik, an AI cost governance platform, and Benchmarkit, a SaaS benchmarking firm , surveyed 396 enterprise organizations across technology, financial services, retail, and manufacturing in April and May 2026. The survey revealed AI costs had a material impact on at least one business decision at 62% of organizations. One in four delayed or canceled an AI initiative outright. One in three deployed emergency spending freezes as bills mounted. And 40% reported that AI spending surprises had escalated to the board. These are not organizations that ignored AI spending. About 95% had assigned formal AI budgets. The problem is that having a budget is not the same as being able to forecast your spending. Only 11% of organizations forecast AI spending within ±10% accuracy. The other 89% are making budget commitments they cannot reliably predict. How AI Costs Broke the Enterprise Budget Model The pattern behind every emergency spending freeze in the report is the same. AI tools piloted in 2024 and 2025 are hitting production in 2026, and production costs dwarf pilot costs in ways most enterprise budgets were never built to absorb. A pilot running 100 users at controlled volume looks nothing like the same tool running 10,000 users at enterprise scale. Token consumption compounds. Infrastructure overhead compounds. Agent retry loops, where a failed agentic task restarts and consumes tokens again, compound. And the governance layer to catch runaway spend was never built because nobody budgeted for it during the pilot. The spending is also coming from places most finance teams are not looking. 98% of engineering organizations use AI coding assistants, yet only 42% include that spending in their AI cost reporting. 68% operate hybrid AI environments spanning cloud and on-premises infrastructure, but fewer than half include on-premises AI resources in formal cost reporting. 43% cited token costs as a top unexpected spending source. The result: 81% of organizations are unable to fully account for their AI costs, and AI has eroded gross margins at four in five enterprises for the second consecutive year. "AI is fundamentally changing how infrastructure is consumed and how costs accumulate," said Sundeep Goel, CEO of Mavvrik. "The organizations succeeding with AI are not necessarily spending less, but they're implementing the governance systems

## 94% of BFSI Firms Use AI to Save Time. Only 19% Track the Returns.

Date: 2026-10-05T11:28:14.483Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/94-of-bfsi-firms-use-ai-to-save-time-only-19-track-the-returns

BFSI firms price risk for a living. They build models to quantify the return on every dollar deployed across loan books, trading desks, and insurance portfolios. They would never deploy capital without a return threshold. They are deploying AI without one. Knolskape's L&D Trends in the BFSI Industry 2026 Report surveyed 68 BFSI organizations globally and found 94.1% use AI to save time and improve work. Only 19.1% track how it impacts revenue. While 60.3% deploy AI for quality and risk, revenue generation sits at 33.8% and new income streams at 27.9%. Fewer than six in ten firms link learning to business outcomes at all, and just 47.1% measure return on investment. BFSI leads every sector in AI adoption. It also leads every sector in the ability to quantify financial outcomes. It applies neither capability to its own AI investment. How AI Became BFSI's Biggest Untracked Expense The 94.1% figure is not evidence of AI maturity. It is evidence of AI access. Saving time and improving work are self-reported efficiency gains with no revenue line. They are the AI equivalent of saying a new filing system improved productivity . The Knolskape data makes the internal prioritization visible. Quality and risk get AI at 60.3%. Customer experience gets it at 45.6%. Revenue generation accounts for 33.8%. New income streams score 27.9%. The deployment pattern tells a consistent story: BFSI firms are using AI to protect what they have, not to build what they could have. That is a defensible short-term choice. It doesn't justify 94.1% adoption without a corresponding revenue measurement framework. Rajiv Jayaraman, Founder and CEO of Knolskape, said that in the BFSI sector, linking learning to measurable outcomes is critical due to regulatory pressures, evolving customer expectations, and high-stakes decisions. The report found that while organizations are directionally aligned on the importance of analytics, the next phase will require a shift from intent to execution. That gap between intent and execution is where the accountability is disappearing. The Human Cost of Unmeasured Disruption The Knolskape findings carry a second metric. One in three BFSI employees is considering leaving because of AI-driven workplace stress. That figure sits alongside data from a UC Berkeley study showing that AI at work does not free up time, it adds responsibilities and extends hours as freed capacity gets filled with expanded scope. Frequent AI users report a 45% burnout rate, compared to 35% among

## Airbnb's CEO Said Anyone Can Build AI Search

Date: 2026-10-05T11:16:52.402Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/airbnbs-ceo-said-anyone-can-build-ai-search

Airbnb launched AI search on September 30, 2026. CEO Brian Chesky had one explanation for why it took this long: "Anyone can vibe-code a search function, but to do something that doesn't kill conversion rate, that's the hard part." Users in the US can now toggle into an AI search mode and describe what they are looking for in plain language, text or voice. The interface generates dynamic filters based on the query rather than presenting a fixed set of categories. Searching for "baby" surfaces cribs, playgrounds, and children's toys automatically. The results include AI-generated descriptions of each property's most relevant features, and saved listings can be compared side by side using AI-powered summaries and review highlights. "We are finally making our first foray into AI search. And what you see is just the very beginning of something," said Brian Chesky, CEO of Airbnb. Airbnb processes approximately $100 billion annually through its platform, which means an AI search experience that reduces purchase intent even marginally has a material financial consequence. Chesky's decision to hold back on AI search until the conversion problem was addressed is the explanation for why Airbnb, a company where AI now writes 60% of new code per a May 2026 disclosure, is only now launching AI-powered property search. The social features announced alongside AI search extend a capability Airbnb began building in 2025. Users can now view an interactive travel map showing where their connections have stayed and where they are going, tapping any trip to see the specific property booked and which experiences were added. Home connections booked can be saved directly to a wishlist. Privacy controls let users determine which trips appear in their network and who can see them, according to the company. The host-facing AI tools arriving later this fall are the most operationally significant disclosure for Airbnb's supply side. A multi-listing calendar allows hosts managing multiple properties to update pricing, availability, cancellation policies, and discounts simultaneously rather than property by property. Dynamic pricing suggests rates based on real-time supply and demand data, with an explanation for each recommendation rather than producing a number without context. An AI earnings dashboard will summarize performance and surface suggested actions. Interactive 3D floor plans will give guests a clearer pre-booking understanding of layout, addressing one of the most common causes of booking hesitation for unfamiliar properties. Airbnb Services continues to expand the post-arrival layer positioning

## DoorDash Built an MCP Connector for Corporate Food Ordering

Date: 2026-10-01T12:10:32.172Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/doordash-built-an-mcp-connector-for-corporate-food-ordering

DoorDash announced on September 30, 2026 that companies can now connect its food delivery platform to their internal AI tools and agents through an MCP connector, letting employees arrange office meals and restock supplies through the workplace tools they already use rather than opening a separate app or website. The connector gives AI agents direct access to four DoorDash capabilities including, searching for items, building a cart, placing an order, and tracking delivery. A company's internal agent, whether a Slack bot, a custom-built enterprise tool, or a third-party AI assistant, can execute the full ordering workflow inside a single conversation, according to the company. An employee can ask an agent to find lunch based on previous orders, set up a recurring daily delivery using an eligible corporate meal benefit, or add items to a group order in a thread that the bot resolves into a single transaction. SpaceXAI, Vercel, Cognition, Tempo, and Mercor are among the early beta partners using the connector. Access is structured at the company level and individual employees sign into their own DoorDash accounts and agree to terms before connecting an agent to act on their behalf, maintaining personal authorization at the transaction layer. "An office lunch can start with a message in Slack and turn into a string of follow-ups: choosing a restaurant, collecting everyone's order, and checking when the food will arrive," DoorDash said in its announcement. "We want to make it easier for the person putting that meal together." Another use case in the announcement is automated office restocking. Companies can connect an agent to their own inventory information, tracking snack levels, drink supplies, or office materials, and have it reorder through DoorDash automatically when stock runs low. The company defines the inventory context and the reorder rules; the connector gives the agent a way to act on them through DoorDash without human intervention at each restock cycle. The connector uses the Model Context Protocol, the same open standard used by Atlassian, DocuSign, ZoomInfo, and Groupon across the enterprise AI integrations covered this quarter. DoorDash is the first major food delivery platform to open an MCP connector for enterprise agentic ordering, extending the protocol from productivity and CRM workflows into corporate food and supply chain operations. DoorDash also announced a separate consumer product on the same day, a text-to-order AI agent accessible via Apple Messages. Users can send prompts such as "order my

## AI Search Is Costco's Best Traffic Source

Date: 2026-09-30T11:18:29.671Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/ai-search-is-costcos-best-traffic-source

On September 29, Costco CFO Gary Millerchip disclosed during their Q4 2026 earnings call that AI search traffic grew triple digits for the second consecutive quarter, and it now converts at a higher rate than every other traffic source Costco tracks. Costco has also never spent a dollar on digital advertising. "While starting from a low base relative to other channels, AI continues to grow in its influence on how our members are searching for products," Millerchip said. "Traffic to our site from AI search grew triple digits for the second consecutive quarter and continues to show the highest conversion rate of all site traffic." AI-referred traffic converting at a higher rate than paid search, email, direct, and social suggests that consumers using AI agents and AI search tools to find products are arriving at Costco with higher purchase intent than any channel Costco has historically measured. The category profile is consistent with Costco's overall e-commerce performance, and AI search-originated sales are led by appliances and consumer electronics. The Costco membership itself is also appearing among the top items originated from AI searches. Millerchip described it as evidence of "strong positive sentiment around the value of our membership coming from these sources," suggesting that AI platforms are organically surfacing the membership value proposition alongside product recommendations. "We don't do any paid digital advertising. AI presents an opportunity in a sort of neutral environment for us to be able to ensure that our value and our quality shows up transparently for our members," said Millerchip. For most retailers, AI search discoverability is a paid media problem, getting products into AI recommendations requires either paid placement, advertising partnerships, or heavy investment in content optimization, according to the company. For Costco, the combination of a curated assortment, buyer-vetted quality reviews, and a brand with strong organic reputation means its products are surfacing in AI results without paid intervention. That structural advantage is what the conversion data reflects. Costco is not treating the AI channel as passive. Millerchip disclosed that the company's technology teams are actively working with Gemini, Anthropic, and OpenAI to clean up and update product data and pages so that Costco's value continues to show up accurately in AI search results. When an analyst asked about agentic commerce positioning, Millerchip pushed back gently on the framing. He described the current moment as AI being used as a search tool rather than

## Robinhood Just Gave 29 Million Retail Investors an AI Trading Agent

Date: 2026-09-30T11:14:07.968Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/robinhood-just-gave-29-million-retail-investors-an-ai-trading-agent

Robinhood launched Robinhood Agents on September 29, opening AI-powered trading agents to all of its approximately 29 million customers at its annual HOOD Summit in Houston. The launch makes Robinhood the first brokerage to offer nontechnical agentic trading at mass retail scale, giving customers without programming knowledge the ability to configure, instruct, and deploy a trading agent directly inside the app, according to the company. Customers choose their preferred model from OpenAI's GPT-6 Luna, GPT-6 Sol, or Anthropic's Opus 4.8, and provide instructions in plain English. Tasks can range from simple executions such as "Buy $200 of Ford stock" to continuous strategies through the Loops feature, which allows an agent to monitor markets and execute trades when defined conditions are met. The agent operates through a dedicated Robinhood trading account, with user-set limits on how much it can trade at a time and a confirmation process that requires user approval before any transaction is executed, according to the company. "We're making Robinhood the best place in the world for active traders by delivering tools once reserved for hedge funds, big banks, and quant firms," said Vlad Tenev, CEO of Robinhood. The launch follows the MCP tool Robinhood released in May 2026, which allowed technically capable users to connect their own agents to its platform. By late September, those early users had already generated nearly 30 million agent transactions per day across 150,000 agentic accounts, providing Robinhood with production data on agent trading behavior before the mass consumer rollout. Robinhood has built three specific guardrails into the agent architecture. Each agent operates through a dedicated account separate from a customer's standard Robinhood account. Customers set explicit trading limits before the agent begins operating. And every proposed transaction requires customer approval before execution, meaning the agent cannot complete a trade autonomously without a human confirmation step. For a limited period, Robinhood is providing agents with free access to data providers including Unusual Whales and the crypto-focused Token Terminal, giving retail customers access to market intelligence sources that have historically been institutional or professional tools. The company's view is that hosting agents do not constitute providing financial advice, describing it as equivalent to a customer asking the internet or a friend. That framing places regulatory responsibility with the customer rather than the broker. The legal landscape around agentic trading is still evolving and no regulatory body has yet issued definitive guidance on where

## Webull Brought Its Trading Platform Into Claude, Perplexity and Grok

Date: 2026-09-30T10:41:46.995Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/webull-brought-its-trading-platform-into-claude-perplexity-and-grok

Webull announced on September 29 that its Cloud MCP server, a fully managed, Webull-hosted infrastructure layer, now supports trade preparation alongside market research and portfolio analysis. The distinction between Cloud MCP and Webull's existing local MCP server is operationally significant. The local server requires users to install software, generate API credentials, and manage access keys, a setup process that restricts the capability to technically proficient users. Cloud MCP eliminates all three requirements. Users connect their existing Webull account through a supported AI platform, authorize the capabilities they want to make available, and begin using Webull's trading tools through a conversational interface without any additional technical configuration. The six capabilities now available through Cloud MCP cover the full pre-trade research workflow, according to the company. Users can query real-time and historical market data across stocks, ETFs, options, futures, crypto, and event contracts; research company fundamentals and regulatory filings; run market screeners; view account balances, positions, and order history; manage watchlists; and have the AI assistant prepare an order based on the research it has conducted — all within the same conversation. "AI is changing how investors access information and interact with their brokerage accounts," said Anthony Denier, Group President and US CEO of Webull. "By bringing trading capabilities to Cloud MCP, we're making it easier for investors to move from research to action through the AI platforms they already use, while keeping them in control of every trade." According to the company, every prepared order requires separate customer confirmation before it can be submitted to the market. The AI assistant completes research, analyzes the position, and prepares the order, but cannot execute it without an explicit human approval at the final step. That confirmation requirement is the same guardrail Robinhood built into its Agents product, announced the same day. Cloud MCP is currently available to eligible Webull users in the United States through Claude, Perplexity, and Grok. ChatGPT availability is pending platform approval. Webull said it is working to expand to additional markets and AI platforms beyond the current US launch.

## AMD Is Acquiring a Physical AI Research Lab for $8.2 Billion

Date: 2026-09-29T12:11:08.254Z
Author: Sachin Mohan
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/amd-is-acquiring-a-physical-ai-research-lab-for-82-billion

AMD announced on September 28, 2026 that it will acquire World Labs, the AI research lab founded by Dr. Fei-Fei Li, in an all-stock deal valued at approximately $8.2 billion. World Labs develops spatial intelligence models that generate, reconstruct, and simulate interactive 3D environments, as well as technology for robotic learning and simulation. Following close, Li will join AMD as Executive Vice President and Chief Scientist, reporting to Chair and CEO Dr. Lisa Su, according to the press release. World Labs' technology enables AI systems to understand and reason about the physical world, generating and simulating 3D environments from text, image, and video inputs rather than operating purely in language or 2D image domains. Applications span robotics, simulation , design, and what AMD describes as physical AI, systems that interact with and operate in real-world environments. "Building the compute platforms for the next generation of AI requires a deep understanding of how models are evolving," said Dr. Lisa Su. "Fei-Fei and the World Labs team bring exceptional research leadership and model expertise. Together, we can use that insight to develop the hardware, software and systems that will power the next generation of AI." The acquisition builds on a working relationship that predates the announcement. World Labs raised $1 billion in a funding round earlier in 2026, with AMD participating as an investor. The two teams had already been collaborating on model training and inference optimization on AMD GPUs. "We began a deep technical partnership with AMD last year," Li wrote on LinkedIn. "As our teams worked together, we realized it would be a natural fit to bring together our AI ecosystem of software and hardware, foundation models, and applications." AMD named three specific AI categories including reasoning, robotics, and simulation, as the domains where compute infrastructure demands are becoming more diverse and where World Labs' spatial intelligence expertise is most directly applicable. Physical AI requires different compute characteristics than language model inference. Higher spatial resolution, more complex simulation workloads, and tighter integration between model training and real-world sensor data. Dr. Fei-Fei Li, known for leading the ImageNet project that catalyzed the deep learning revolution in computer vision and for co-directing the Stanford Human-Centered AI Institute, will join AMD as Executive Vice President and Chief Scientist, reporting to Su. The appointment gives AMD a researcher whose career has shaped the trajectory of AI development at one of the most critical junctures

## Shopify Just Let Browser-Based AI Agents Complete Purchases

Date: 2026-09-29T11:44:29.543Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/shopify-just-let-browser-based-ai-agents-complete-purchases

On September 28, 2026, Shopify announced that browser-based AI agents can now complete purchases on Shopify merchant sites. With the buyer’s authorization, these agents can navigate checkout screens, update order details and submit transactions. The technical change introduces three new tools through Shopify's WebMCP support. The get_checkout tool allows an agent to read the current state of the checkout screen. The update_checkout tool allows it to modify details, changing a delivery address, selecting a shipping option, adjusting quantities. The complete_checkout tool submits the transaction after the buyer authorizes it. All three operate through structured APIs designed specifically for agents, rather than requiring the agent to interpret screenshots or scrape HTML pages built for human navigation, according to the company. "If your agent is operating in the buyer's browser, use WebMCP tools provided on storefront and checkout to efficiently complete order placement, instead of navigating HTML built for humans," said Gil Greenberg, Staff Product Manager for Checkout at Shopify. "These WebMCP tools provide structured and efficient APIs, purposely designed via UCP, to ensure accurate commerce facts, required disclosures, and handoff requirements." The feature is rolling out to all eligible Shopify merchants globally. The WebMCP checkout update sits on top of Shopify's Universal Commerce Protocol, a shared language that both WebMCP and Shopify's hosted MCP server speak. WebMCP handles agents operating inside the buyer's browser, according to the company. The hosted MCP server handles server-to-server agent interactions. Both routes lead to the same checkout and Shop Pay infrastructure, meaning agents built on different architectures can complete purchases through the same Shopify commerce layer. Two named agent partnerships went live alongside the announcement. Meta's Muse, whose Shopify partnership was announced the previous week, can now complete full agentic checkout through Shop Pay across all Shopify stores. Instinct, a personal shopping agent, announced its Shopify integration on the same day. Instinct users can search Shopify's merchant network for products, check live prices, sizes, and availability, and complete checkout with Shop Pay using saved payment and shipping details, all within a single conversation. Amazon blocked Meta's Muse from conducting purchases on its platform the same week Shopify opened its checkout to browser agents. Adidas has also blocked agent-initiated purchases. Shopify's decision to build the infrastructure layer for agentic commerce reflects the same strategic logic behind its Meta partnership: AI agents make it easier to find and purchase products, and Shopify's merchant network and Shop Pay

## Mastercard Launched an AI-Powered B2B Analytics Platform

Date: 2026-09-29T10:54:22.907Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/mastercard-launched-an-ai-powered-b2b-analytics-platform

Mastercard launched Advanced B2B Analytics on September 28, 2026 to address a gap that sits inside one of the largest payment markets in the world. Banks and corporate buyers currently have no systematic, data-driven way to identify which suppliers in their payables portfolio are most likely to accept card payments. So commercial card programs expand slowly, through broad outreach, rather than by targeting the highest-probability conversions first, according to the company. The platform addresses that gap with AI-powered acceptance propensity scoring, analyzing payment and accounts payable data to rank suppliers by their likelihood of accepting card payments and surface the highest-value conversion opportunities first. Interactive dashboards translate the scoring into actionable analytics that corporate clients can use to optimize payment flows, reduce reliance on manual processes, and improve working capital visibility across their supplier networks. "Businesses today are sitting on vast amounts of payment data, but too often it remains fragmented and underutilized," said Marc Pettican, Global Head of Corporate Solutions at Mastercard. "With Advanced B2B Analytics, we're combining powerful insights with our deep commercial payments expertise to help issuers identify where card acceptance can deliver the greatest value, engage suppliers more effectively and drive measurable growth." Absa Group and Emirates NBD are the first financial institutions named as deploying the platform. Both are using it to identify supplier acceptance opportunities for their corporate clients and accelerate virtual card adoption across their commercial payments portfolios. Mastercard estimates the addressable B2B payments market at approximately $80 trillion globally, the majority of which still flows through checks, ACH, and wire transfers rather than commercial cards. According to the company, the friction that keeps that conversion from happening faster is the same friction the platform targets: fragmented data, manual supplier outreach, and the absence of a systematic way to prioritize where commercial card programs can expand most efficiently. Advanced B2B Analytics is combined with consulting from the Mastercard Advisors and Transformation team, giving financial institutions access to both the data layer and the advisory capability to act on it. Mastercard described the platform as the first foundational B2B capability in a broader AI-powered commercial payments infrastructure, one it plans to extend to the supplier and acquirer side of the ecosystem in subsequent releases, progressively embedding intelligence across the full commercial payments journey. "Our partnership with Mastercard gives clients a smarter way to understand their payment ecosystems, unlock value across their supplier networks and

## Cognex Launched an AI-Powered Wafer Reader for Semiconductor Fabs

Date: 2026-09-29T04:42:28.307Z
Author: Sachin Mohan
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/cognex-launched-an-ai-powered-wafer-reader-for-semiconductor-fabs

As semiconductor manufacturers scale capacity to meet AI infrastructure demand, wafer identification is emerging as a production bottleneck. Degraded marks, varied substrate materials, and non-standard fonts create read failures that trigger manual intervention and reduce equipment utilization. Cognex launched the In-Sight 1750 Series on September 23, 2026 to address that bottleneck with an architecture that adds Edge AI OCR alongside established rule-based vision rather than replacing it. Semiconductor manufacturing requires a level of process predictability and auditability that makes replacing proven systems with AI-only approaches operationally risky, according to the press release. The In-Sight 1750 applies Edge AI OCR specifically to the cases where traditional rule-based OCR struggles like degraded marks caused by the fabrication process, non-standard fonts on specialty substrates, and challenging materials including silicon carbide, gallium nitride, sapphire, and glass, while keeping rule-based systems handling the standard reads they have been proven to perform reliably. "Investment in AI infrastructure is driving a new wave of expansion across the semiconductor industry, and chipmakers are being asked to bring new capacity online faster than ever while maintaining the highest quality standards," said Matt Moschner, President and CEO of Cognex. "The In-Sight 1750 combines our latest AI technology with decades of semiconductor expertise to help manufacturers scale production without sacrificing quality or traceability." Even a small number of missed wafer reads in a semiconductor fabrication line creates downstream consequences that compound across a production run. In a manufacturing environment where capacity expansion is being driven by AI infrastructure demand, the operational cost of wafer identification failures grows proportionally with production volume. According to the company, the In-Sight 1750 is built with existing In-Sight 1740 users in mind. It supports job conversion from the previous generation and is designed to allow manufacturers to transition to Edge AI OCR capabilities without extensive equipment redesign or process requalification, reducing the adoption barrier that has historically slowed machine vision upgrades in semiconductor environments where any process change requires validation. Purpose-built illumination is a core component of the system rather than an afterthought. Wafer marks are often degraded by the fabrication process itself, which means the imaging conditions under which OCR operates directly determine read reliability. The In-Sight 1750's illumination is designed specifically for the contrast characteristics of semiconductor substrate materials rather than adapted from general-purpose machine vision hardware. Built-in read logging provides the audit trail that semiconductor quality systems require, recording every read result for

## AI Just Added $942 Million to US Healthcare Costs

Date: 2026-09-28T10:48:16.994Z
Author: Sachin Mohan
Desk: Pharma
URL: https://aimmediahouse.com/pharma/ai-just-added-942-million-to-us-healthcare-costs

The Blue Cross Blue Shield Association published an analysis on September 24, 2026, finding that hospitals' AI coding tools generated $942 million in additional healthcare spending over two years, with no evidence the patients involved received different treatment.. The BCBSA found a sharp increase in patients being documented as having complex conditions and no evidence that those patients received different treatment as a result. The mechanism the BCBSA identified is operationally specific. AI coding tools used by hospitals scan patient records and suggest that physicians document the severity of existing conditions, flagging, for example, that a patient's kidney disease meets the threshold for a more severe classification, or that a test ordered during a visit implies a secondary diagnosis that would otherwise go unrecorded. By adding conditions like anemia or low sodium alongside a primary diagnosis, hospitals reclassify patients as having more complex conditions. The average additional payment per case where that reclassification occurred was nearly $12,000, according to the analysis. McLaren Health Care, a Michigan health system, offered the most specific named example in the New York Times coverage . CFO Dave Mazurkiewicz confirmed that McLaren uses AI technology from SmarterDx, which takes a cut of the additional revenues it generates, and has increased revenue by $1 million per month through its adoption. "All the insurers are using AI to scan our charts to look for claims to deny," Mazurkiewicz told the Times. "For the same reason, we're looking at the same charts today." The Bilateral Escalation and Why It Is Getting Worse The BCBSA finding describes one side of a two-sided dynamic. Hospitals are using AI to document more and collect more. Insurers are simultaneously using AI to scan claims and deny more. Both sides are escalating, and the structural reason the escalation is accelerating is that the cost of each additional round has dropped. Caroline Pearson, Executive Director of the Peterson Health Technology Institute, identified the dynamic precisely. Both sides are willing to go additional rounds in the billing dispute "because it's cheap," she told the New York Times. When human reviewers had to manually assess every claim and every appeal, the cost of escalation constrained how aggressively either side could fight. AI removes that constraint. A hospital can now file AI-assisted appeals at a volume and speed that would have been operationally impossible without the technology. An insurer can now scan every claim against a database

## Stitch Fix Named AI Investment as a Reason for EBITDA Declining

Date: 2026-09-25T11:25:44.846Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/stitch-fix-named-ai-investment-as-a-reason-for-ebitda-declining

Stitch Fix is spending more on AI in fiscal 2027 than it did in fiscal 2026, and CFO David Aufderhaar confirmed on the September 23 earnings call that this investment is one of the two primary reasons adjusted EBITDA guidance has fallen from $51.9 million to a midpoint of $34.5 million for the year ahead. "Excluding these additional investments, the EBITDA margin guide would have been much more consistent with last year's margin at the 4% range," Aufderhaar said. The AI investment is concentrated on Vision, Stitch Fix's style visualization platform, and on expanding the operational AI that has already produced measurable results across styling, fulfillment, and customer service. AI agents are now handling enough routine customer service volume that Stitch Fix has more than doubled the share of chats resolved without requiring additional human support. AI is improving daily labor planning and real-time staff scheduling in fulfillment centers. Generative AI is supporting stylists in item recommendations so they can redirect time to higher-touch client interactions, according to the company. Vision is the commercial bet underlying all of it. Stitch Fix has shared more than 22 million Vision images with clients. The recently launched "See It on Me" feature allows shoppers to see outfit ideas rendered on their own likeness, extending Vision from a browsing tool into a personalized try-on experience. CEO Matt Baer said adoption has exceeded expectations and the company continues to see a significant lift in 90-day Freestyle spend from clients who engage with it. An analyst from Mizuho asked directly whether the rise of Meta's Muse personal AI agent, an autonomous shopping assistant that can discover and purchase products on a user's behalf, presents a competitive risk to Stitch Fix's model. Baer argued that the quality of an AI experience depends on the quality of the underlying data, and that Stitch Fix's 15 years of granular client preference data, combined with human stylist relationships that deepen with every Fix, constitutes a competitive advantage that generic AI agents cannot replicate for high-nuance purchases. "If I want to ensure my pantry is perpetually replenished with paper towels, that's something I would happily outsource to my agentic shopper," Baer said. "If I want to ensure that I'm dressing style and trend appropriate with clothes that fit me, that align to my budget, and that will give me confidence when I walk out the door, there's a lot more nuance

## Accenture Invested in a Platform That Maps How Work Actually Happens

Date: 2026-09-25T05:30:35.444Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/accenture-invested-in-a-platform-that-maps-how-work-actually-happens

Accenture's Pulse of Change survey found that 82% of C-suite leaders are increasing AI investment in 2026. Only 23% say it is delivering widespread sustained business value, down from 32% earlier in the year. The investment in Within, announced on September 23, 2026, is a direct response to the gap between those two numbers. The reason AI fails to scale across most enterprise deployments is documented. The processes that AI agents are asked to automate are not the processes that actually run the business, according to the press release. Official process documentation describes how work is supposed to happen. What employees actually do is rarely written down anywhere. AI deployed on the official process fails because the official process is not what the organization is actually running. According to the press release, Within's platform captures what the official documentation misses. It tracks how employees work across all applications in real time and records the undocumented interactions that shape how work actually flows. That information is compiled continuously into the Work Brain: a live context layer that AI agents can draw on when executing tasks rather than working from process maps that do not reflect operational reality. "One of the most persistent challenges in AI transformation is that organizations often don't have a clear picture of the starting point," said Jason Dess, Accenture's Industry and Process Reinvention Lead. "Without that understanding, it's difficult to scale AI beyond isolated use cases." Blackbaud, a software company serving the social impact sector, is the named customer in the announcement. EVP and Chief Commercial Officer David Benjamin described what Within surfaced as the difference between the process on the diagram and the process that was actually running. "You can't reinvent what you can't see," Benjamin said. "Accenture and Within gave us that visibility, and the plan to act on it." The architecture Within uses to capture that process data is a private, sovereign model, designed specifically so that the operational intelligence it generates stays within the client's own infrastructure rather than flowing through shared cloud environments. That design makes it deployable in regulated industries including financial services, healthcare, and government, where process data is sensitive and standard SaaS data handling is not acceptable. "AI transformation works best when there's a system of record of work," said Andrew Antos, CEO and Co-Founder of Within. "People and agents working in harmony is the future of work."

## Retail's Most Expensive Problem Right Now Is AI Search

Date: 2026-09-24T14:46:49.885Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/retails-most-expensive-problem-right-now-is-ai-search

During MillerKnoll’s Q1 FY2027 earnings call on September 22, Debbie Propst, President of Global Retail, named Google's AI Mode specifically as a factor driving up the company's digital advertising costs. The company disclosed that it was redirecting the budget into direct mail to compensate. "We are definitely seeing increased digital advertising costs," Propst told investors. "In August, we leaned more heavily into our direct mail distribution.” She informed that the company will continue through the remainder of the year, with the upcoming midterm elections likely to drive digital marketing costs more as well. AI search was named as a specific and material cost driver. The Cost Structure Behind the Comment AI Overviews, Google's AI-generated answer summaries at the top of search results, have reduced organic visits by 61% and paid visits by 68% since mid-2024, according to Bounteous analysis of queries with informational intent. As AI search platforms resolve user questions on-page, fewer users click through to brand websites, and the clicks that do happen are increasingly concentrated among brands cited directly within the AI-generated answer. Seer Interactive found that when a brand is cited in an AI Overview, paid click-through rate is 91% higher than when it is not cited, confirming that AI search visibility and paid performance are now directly connected rather than independent channels. The cost escalation is not confined to search . Meta's average price per ad rose 12% year over year in both Q1 and Q2 2026, up from 9% full-year 2025, while ad impressions grew 14% in Q2. Triple Whale, measuring across approximately 35,000 brands for full-year 2025, puts the median eCommerce CPM on Meta at $14.19, up 20% year over year. The same supply and demand dynamics driving cost increases in traditional social advertising are now compounding in AI search, where brands are competing for fewer available impression slots in a more concentrated discovery environment. Gartner forecast in February 2024 that traditional search engine volume would fall 25% by 2026 as AI chatbots and virtual agents resolve queries without generating click-through traffic. Adobe data published in January 2026 shows AI-driven referral traffic to US retail sites surged 693% year over year during the 2025 holiday season. The volume is moving to AI channels faster than the advertising infrastructure to reach users within those channels has been built out. The Conversion Paradox and the Measurement Gap AI-referred traffic to US retail sites converted 54%

## Oracle Health Announced Five AI Tools for Revenue Cycle Management

Date: 2026-09-24T13:43:45.223Z
Author: Sachin Mohan
Desk: Pharma
URL: https://aimmediahouse.com/pharma/oracle-health-announced-five-ai-tools-for-revenue-cycle-management

Most revenue cycle denials trace back to issues that existed before the claim was filed including missing documentation, failed authorizations, incorrect charges, or coding errors. Oracle Health announced five AI capabilities on September 23, designed to identify those issues. The five tools cover the five most common denial origination points. Prior authorization, checks coverage, retrieves documentation requirements, pre-fills the relevant fields, attaches supporting evidence, and coordinates payer interactions before the claim is submitted. Clinical documentation quality integrity reviews reimbursement-related documentation and surfaces recommendations to close gaps before coding. Charge capture and integrity analyzes clinical context, modifiers, and service details within charge review workflows to limit the coding errors and omissions that produce revenue leakage. Medical coding for professional fees recommends codes from clinical documentation within existing workflows. Appeal management identifies payment variances in remittances received and automates the creation of appeal packets when a decision to contest a denial is made. "AI gives us an opportunity to prevent revenue cycle problems before they lead to denials and delayed payments," said Seema Verma, EVP and General Manager, Oracle Health and Life Sciences. "Oracle Health is delivering a comprehensive revenue cycle solution that brings AI across the front, middle and back office, connecting clinical and financial workflows from the first patient interaction through payment." According to the company, the design principle across all five tools is upstream intervention. Revenue cycle teams currently spend significant time correcting problems that were created earlier and that correction work compounds as claims move further through the billing cycle. Oracle Health's stated approach embeds AI at the point where information is first entered rather than at the point where a denial arrives. The tools are embedded natively in existing Oracle Health revenue cycle solutions, covering patient access, clinical documentation, patient accounting, contract intelligence, payment operations, enterprise analytics, and financial management, so healthcare organizations can adopt the AI capabilities without replacing their existing infrastructure. Oracle Health also disclosed plans to connect reimbursement workflows with enterprise financial operations through Oracle Fusion Cloud Applications, including revenue accounting, treasury management, and financial reconciliation, extending the scope of the AI initiative beyond revenue cycle into enterprise financial management. The announcement follows Seema Verma's December 2025 preview of a broader set of AI agents planned for 2026 across revenue cycle, nursing, and clinical operations.

## YouTube Embedded AI Into Every Stage of the Creator Workflow

Date: 2026-09-24T12:38:53.742Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/youtube-embedded-ai-into-every-stage-of-the-creator-workflow

YouTube revealed a set of AI tools at its Made on YouTube event on September 23, covering video production, channel management, brand partnerships, and commerce across the creator workflow. Ask Studio is the most significant new addition, according to the company. It provides channel performance summaries, viewer comment analysis, and personalized strategic guidance, and also helps creators respond to brand briefs and develop pitch materials within YouTube Studio. It also assists with brand partnerships, helping creators respond to briefs and develop pitch materials, which moves YouTube directly into the commercial relationship between creators and advertisers rather than simply facilitating the transaction. YouTube Studio is also adding A/B testing for up to three video cuts, feedback on early drafts before a video is published, and channel-matched thumbnail generation that fits a creator's existing visual style rather than producing generic options. The Gemini-powered conversational editing tool, available in Shorts and the YouTube Create app, allows creators to use natural language prompts to trim pauses, rearrange clips, and improve pacing rather than working through manual editing timelines. Manual editing remains available for creators who want granular control. "Our vision is for Ask Studio to become the ultimate creative partner for every Creator, a trusted companion that Creators turn to first," YouTube said in its announcement. YouTube Shopping's affiliate program, already at 1.3 million enrolled creators, is also expanding to 35 countries by the end of 2026. Creators in India and Brazil will be able to tag Amazon products in their videos, extending Amazon's agentic commerce reach into YouTube's creator ecosystem, according to the company. Product tags will automatically display localized offers based on where a viewer is watching, making the shopping layer market-specific rather than applying a single global price and availability. Ask YouTube's expansion into shopping queries adds a product discovery dimension to YouTube's search layer. Users will be able to ask about products, receive organized recommendations, and see comparison tables, with voice commands available for TV users. The feature connects YouTube's existing strength in product review and demonstration content to a direct purchase pathway rather than requiring viewers to leave the platform to complete research. YouTube said its Shopping affiliate expansion and Ask YouTube shopping features together reflect a move toward integrating commerce across both the creator and viewer experience, connecting the moment of product discovery in a video to the moment of purchase without routing through a separate e-commerce surface.

## Amazon's Seller Assistant Now Runs 24/7 Without the Seller

Date: 2026-09-24T11:36:39.666Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/amazons-seller-assistant-now-runs-247-without-the-seller

Seller Assistant started in 2023 as a generative AI companion that answered questions and surfaced insights. On September 23, 2026, Amazon announced the version that works whether a seller is logged in or not. With the new workflows capability, Sellers can describe what they need in plain language, set their guardrails, and the workflow runs continuously, according to the company. It monitors conditions, interprets changes, and responds dynamically without requiring the seller to ask. Every action is logged with a complete audit trail. Sellers choose upfront whether the workflow surfaces recommendations for their review or takes actions on their behalf, keeping human control at the design stage rather than requiring intervention at every execution. According to the company, Seller Assistant also now carries persistent memory. Pricing patterns, inventory cycles, and growth goals are stored and follow the seller across Seller Central, Amazon Quick, and Claude, so context is never lost when switching between tools. The system is built on Amazon Bedrock with Anthropic's Claude models, with seller data staying within Amazon's infrastructure and never shared outside it. "Amazon sellers make dozens of decisions a day on pricing, inventory, and ads," said Eric Burns, Field CTO at Anthropic. "Amazon Seller Assistant already uses Claude models to help sellers make informed, strategic decisions, and they accept its recommendations more than 90% of the time." The selling partner plugin is the second major disclosure in the announcement. Ninety percent of Amazon's selling partners already use third-party AI tools to run their businesses, and the plugin is designed to bring Seller Assistant's intelligence into those tools rather than requiring sellers to switch to Amazon's interface. The plugin connects each seller's Amazon data including listing contributions, real-time performance metrics, inventory levels, and sales analytics, directly to their preferred AI agent. It does not stop at data access. The agent can act on the seller's Amazon business through the plugin, the same way Seller Assistant does inside Seller Central, with human approval required on every action. Connection takes approximately 60 seconds with no coding required. According to the company, the plugin launches first in Amazon Quick, Amazon's general-purpose AI companion for business, and in beta with Anthropic's Claude. Amazon Quick Plus is available free to every primary account holder globally through December 31, 2026, with primary account holders able to designate two additional co-workers for free access. International expansion of the plugin is planned to follow

## Kroger Wants AI Agents to Handle Your Weekly Grocery Shopping

Date: 2026-09-23T11:09:47.102Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/kroger-wants-ai-agents-to-handle-your-weekly-grocery-shopping

At GroceryShop 2026 on September 22, Kroger described an AI strategy anchored in two things most grocery competitors cannot easily replicate. A decades-long loyalty data foundation and a governance architecture that calibrates autonomy by risk rather than applying it uniformly. Yael Cosset, EVP and Chief Digital Officer, described the strategy in three stages. Assisted AI supports human decisions without replacing them. Augmented AI enhances human capabilities by surfacing insights and automating repetitive tasks. Autonomous AI acts independently within clearly defined governance boundaries. The critical qualifier is that autonomy is not the default destination, it is the appropriate destination for some use cases and the wrong one for others, said the company. "The goal is not autonomy everywhere," Cosset said. "The goal is using the right level of AI based on risk, repeatability and the need for human judgement. Different cases will sit at different stages, and strong governance remains at the center." The framework governs three current deployments . The AI Shopping Assistant helps customers plan meals, find recipes, and build carts around budget, dietary needs, and preferences, sitting at the assisted stage where the customer retains decision authority. Conversational search extends the same capability to product discovery and meal planning. AI tools for frontline associates filter the dozens of daily notifications that store teams receive and surface only the most important tasks, an augmentation use case designed to reduce noise rather than replace judgment. Kroger's argument for why its AI strategy is structurally differentiated from other grocery retailers is the data foundation underneath it. The company's loyalty program has generated decades of customer shopping data that its proprietary models are trained on, according to the company. Those models span demand forecasting, operational optimization, personalization, and generative AI capabilities, and are combined with third-party frontier models through shared infrastructure that includes data services, agentic capabilities, and governance guardrails. "This isn't new for Kroger," Cosset said. "We've long been using data science and artificial intelligence to serve our customers by precisely forecasting demand, connecting customers with the most relevant savings opportunities and helping them find and get the products they need." The most forward-looking element of the GroceryShop presentation was Cosset's description of grocery as a future managed service. In that model, customers share goals and AI agents handle the planning, shopping, and fulfillment on their behalf rather than requiring customers to formulate a weekly shopping list and navigate a retail

## MVB Bank Is Automating Fintech Partner Risk Reviews With AI

Date: 2026-09-23T09:24:00.175Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/mvb-bank-is-automating-fintech-partner-risk-reviews-with-ai

MVB Bank's fintech partner business creates a specific compliance challenge. The bank powers payments, card issuance, sponsorship lending, and online gaming programs for fintech companies nationwide. Each partnership requires its own due diligence, ongoing monitoring, and marketing compliance review. At scale, those reviews accumulate faster than manual processes can handle them consistently. The partnership with Kobalt Labs announced on September 22 is designed to close that gap with AI-driven automation rather than additional headcount. Kobalt's platform automates the workflows that third-party risk management requires including vendor reviews, policy analysis against current regulatory expectations, security assessments, and marketing compliance checks, and is designed to produce risk assessments that are not just faster but consistent and defensible across every partner review. Consistency is the word that appears most often in the regulatory guidance that governs bank-fintech partnerships, according to the press release. A risk assessment that is thorough but produced differently each time does not satisfy the documentation standard that regulators apply when reviewing a bank's third-party oversight program. "If you want to show your regulators you're doing it right, you have to add consistency and scale with technology," said Kalyani Ramadurgam, CEO and Co-founder of Kobalt Labs. The partnership was introduced by Klaros Group, a regulatory consulting firm that works with both MVB and Kobalt. According to the press release, regulatory consultants are increasingly functioning as the connective tissue between banks that need to modernize compliance workflows and AI platforms that can automate them, because the consultants understand both the regulatory requirement and the technology landscape simultaneously. "Third-party risk management is one of the areas where banks are under real pressure to do more, faster, and with greater consistency," said Adam Shapiro, Co-founder and Partner at Klaros Group. "Kobalt brings a market-leading AI platform to that challenge, and when combined with Klaros' deep regulatory expertise, it creates a powerful solution for institutions like MVB." Guidance from the OCC, Federal Reserve, and FDIC on third-party risk management, updated in 2023 and increasingly enforced in 2025 and 2026, requires banks to demonstrate documented, consistent oversight of every third-party relationship, including fintech partners. "MVB is committed to being at the forefront of technological innovation," said Larry F. Mazza, CEO and President of MVB Financial. "We share a strong commitment to effective risk oversight and management while meeting regulatory requirements, and we're confident this partnership will help us do that at scale."

## University Hospital at Downstate Deploys Altera’s AI Scribe Across Clinical Care

Date: 2026-09-22T19:19:02.928Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/university-hospital-at-downstate-deploys-alteras-ai-scribe-across-clinical-care

University Hospital at Downstate (UHD), the academic medical center of SUNY Downstate Health Sciences University, has deployed Altera Digital Health’s Sunrise Thread AI to automate clinical documentation during patient visits. The ambient AI scribe captures patient-provider conversations in real time and generates a structured clinical note inside Altera’s Sunrise electronic health record (EHR). Since the pilot began in May, 50 providers across family medicine, internal medicine, cardiology and the emergency department have used the system in more than 900 patient encounters, Altera said. The deployment comes as health systems expand the use of ambient AI to reduce the documentation burden on clinicians. Other health systems have also been moving AI scribes into clinical workflows. UHD serves Brooklyn and is the only academic medical center in the borough. The hospital has worked with Altera as its EHR partner for nearly 20 years. AI Scribe Built Into Existing EHR Sunrise Thread AI is designed to operate within the existing Sunrise environment rather than requiring clinicians to switch between separate applications. Providers can record conversations using a mobile device, while the resulting note can be reviewed and approved in the EHR. Altera introduced Sunrise Thread AI in 2025 as an ambient documentation tool designed to capture clinical conversations and convert them into structured notes. The company positioned the technology as part of its effort to reduce the time clinicians spend on manual documentation. The approach follows a wider push to bring AI directly into electronic health records. Sutter Health, for example, has also incorporated AI capabilities into its clinical workflows. Dilip Nath, Chief Information Officer at SUNY Downstate Health Sciences University, said the early response from clinicians had been positive and that demand for the technology remained high. UHD is tracking the system's usability, transcription and interpretation quality, documentation workflow and potential improvements. Altera and the hospital are also developing customized templates for inpatient and emergency department physicians based on early feedback. Downstate Expands Use Across Clinical Teams Mafuzur Rahman, Vice Chairman and Director for Inpatient Services and Chief Medical Informatics Officer at SUNY Downstate Health Sciences University, said the system had given clinicians more time while improving the completeness of documentation. The hospital also expects the technology to help medical trainees who rotate through different clinical settings. UHD has nearly 1,000 students, interns, residents and fellows who have varying levels of familiarity with its EHR. Patient consent remains part of the implementation.

## AbbVie Partners With Iambic to Accelerate AI-Driven Drug Discovery

Date: 2026-09-22T19:09:02.079Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/abbvie-partners-with-iambic-to-accelerate-ai-driven-drug-discovery

AbbVie has entered a multi-year collaboration with clinical-stage life sciences and technology company Iambic to accelerate the discovery and development of small-molecule therapies across immunology, neuroscience and oncology. Under the agreement, Iambic will apply its AI-driven drug discovery platform to programs with potential to become first-in-class or best-in-class medicines. AbbVie will contribute its therapeutic-area and drug discovery expertise, while Iambic will provide its AI technologies and automated chemistry and biology capabilities. The financial terms include an upfront payment to Iambic, along with potential success-based milestone payments and tiered royalties on net sales of products generated through the collaboration. The companies did not disclose the value of the upfront payment or potential milestones. The partnership was announced on 09/21, the same day Iambic unveiled the latest version of its AI model, Enchant v3. Iambic Brings Enchant v3 Into AbbVie Collaboration Enchant v3 is a multimodal transformer model that combines biomedical data from across the drug discovery process. Iambic said the model was trained using 41 billion parameters across more than 6,000 molecular properties and 16 biomedical data modalities. It is designed to test multiple drug-design hypotheses in parallel and help researchers evaluate drug properties earlier in the discovery process. AbbVie said the collaboration will use Enchant v3 alongside Iambic's NeuralPLexer technology. NeuralPLexer is designed to predict protein-ligand complexes, while Enchant focuses on multimodal prediction across drug discovery and development. In traditional small-molecule discovery, researchers evaluate numerous properties for individual candidates. Iambic's platform is designed to evaluate multiple properties simultaneously, which the companies said could help researchers make more precise decisions about which molecules to pursue. The approach combines computational prediction with laboratory work. That fits with the industry's increasing use of AI to guide drug discovery while experimental testing remains necessary to validate potential therapies, a dynamic explored in AI drug discovery's changing role in pharmaceutical research. Jonathon Sedgwick, Senior Vice President and Global Head of Discovery Research at AbbVie, said combining Iambic's platform with AbbVie's drug-hunting expertise could help the company make more precise decisions about the molecules it pursues. The goal, he said, was to design better molecules more efficiently and improve the probability of success. Iambic Expands Its Pharma Partnerships The AbbVie agreement adds to a series of pharmaceutical collaborations for Iambic this year. In June, Bayer entered a collaboration with Iambic focused on small-molecule discovery for hard-to-drug targets. The agreement gives Bayer access to Iambic's Enchant and

## Disney's First CTO Will Oversee AI, Data and Engineering Across the Entire Company

Date: 2026-09-22T11:41:21.414Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/disneys-first-cto-will-oversee-ai-data-and-engineering-across-the-entire-company

A year ago, Disney's lawyers sent Character.AI a cease-and-desist letter demanding it stop using Disney characters without authorization. On September 18, 2026, Disney hired Character.AI's CEO as its first-ever Chief Technology Officer. Karandeep Anand joins Disney as Senior Executive Vice President and CTO effective October 2, reporting directly to CEO Josh D'Amaro in a role that did not exist before this appointment. A number of Character.AI's technical team are expected to follow him. The mandate covers enterprise technology, infrastructure, data and AI platforms, product, and engineering across Disney's full business, according to the company. D'Amaro, who took over as CEO in March 2026 replacing Bob Iger, framed the appointment in terms of the three priorities he has established since taking the role: great storytelling as the North Star, technology in service of creativity, and operating as One Disney. At Disney's Q3 2026 earnings call in August, he told investors the company is using AI across its studios and the full enterprise, for production pipeline efficiency, film rendering and noise reduction, expanding the number of films available in 3D, and improving Disney+ personalization and recommendation. He also named data unification as an enterprise-wide priority, describing how large companies built through acquisitions often end up with disparate datasets that do not communicate across business units. "Karandeep brings a rare mix of experience across infrastructure, consumer technology and AI," D'Amaro said. "He is an expert builder and strategist, technical and product-minded, with sharp instincts for what fans actually want." In September 2025, Disney sent Character.AI a cease-and-desist letter alleging copyright infringement of characters including Elsa, Moana, Peter Parker, and Darth Vader. Character.AI removed those characters. The dispute did not prevent Anand from being appointed, and the decision to bring him in alongside members of his technical team signals that Disney's priority in hiring him is his AI platform-building experience rather than his association with the copyright dispute. Anand led Character.AI through a period of rapid growth, building one of the most engaged consumer AI audiences in the world while making user trust and safety a structural priority as the platform scaled. Before Character.AI he served as President and Chief Product Officer at Brex, the financial technology company, and as VP of Ads and Business Products at Meta's Facebook. He spent 15 years at Microsoft, where he was involved in building the Azure platform. His academic background is a Computer Science degree from the

## Amazon Is Deploying AI Glasses to 20,000 Delivery Drivers

Date: 2026-09-22T11:30:42.259Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/amazon-is-deploying-ai-glasses-to-20000-delivery-drivers

Amazon disclosed two specific AI tools at its Ignite Live conference in Las Vegas on September 21. One has been building a parking and access database since 2024 and the other puts navigation instructions in a driver's line of sight. Wellspring is Amazon's generative AI mapping tool for last-mile delivery logistics. According to the company, it has identified 202 million parking locations, mapped 2.8 million building entrances, and cataloged 85,000 mailrooms and lockers since 2024. This is a proprietary dataset that Amazon is extending with AI routing capabilities designed to account for conditions that standard navigation tools do not address specifically. Those conditions include reduced visibility after sunset and the additional time required to locate parking in dense urban environments, factors that affect delivery speed and driver safety in ways that GPS alone cannot resolve. Amazon's Smart Delivery Glasses extend the same operational AI logic into wearable hardware. The glasses display navigation suggestions and delivery details in the driver's line of sight, allowing hands to stay free during the final steps of each delivery, according to the company. New features being rolled out include hands-free scanning and proactive address verification, reducing the time drivers spend checking delivery details on a handheld device at each stop. Amazon expects more than 20,000 Smart Delivery Glasses to be in the field by the end of 2027. "We're focused on using hardware and AI-backed software systems to help DSP owners run stronger operations and improve the delivery experience," said Beryl Tomay, VP of Transportation at Amazon. The AI tool deployments sit within a broader safety investment that Amazon quantified at the conference. Serious vehicle crashes across its delivery network dropped more than 23% last year, a result Amazon attributes in part to the more than $543 million it has invested in safety initiatives, programs, and technology for DSP owners since 2022. Surround-view cameras that identify vehicles, cyclists, and pedestrians from multiple directions are being installed in 50% of Rivian electric delivery vehicles by end of 2026, with dynamic hazard alerts and automatic route adjustments also being added to the Amazon Delivery App. The $1.9 billion 2027 DSP investment brings total program investment to $21.7 billion over eight years, and is expected to push national average pay for Delivery Associates to nearly $24 per hour, up approximately $1 from current levels. Over the past three years, Amazon said its investments have helped DSP owners increase

## Meta's Muse Can Now Buy From Shopify Stores. Amazon Blocked It the Same Weekend.

Date: 2026-09-22T11:22:44.268Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/metas-muse-can-now-buy-from-shopify-stores-amazon-blocked-it-the-same-weekend

Meta's Muse personal AI agent can now complete purchases on behalf of users across all Shopify-powered stores, announced on September 21, making it the first major personal AI agent to integrate agentic checkout at the scale of Shopify's merchant network. The integration works through Shop Pay, Shopify's accelerated checkout service that stores payment and shipping details for faster transactions, meaning a Muse user can ask the agent to find and buy a product, and Muse completes the transaction through Shop Pay without the user navigating to a retailer's website or entering payment details. Shopify CEO Tobi Lütke described it on X as enabling "agentic checkout with Shop Pay on all Shopify stores," covering the full breadth of Shopify's merchant base rather than a subset of participating retailers. The same weekend, Amazon blocked Muse from conducting shopping through its platform. "We think it's fairly straightforward that third-party applications that offer to make purchases on behalf of customers from other businesses should operate openly and respect service-provider decisions about whether or not to participate," an Amazon spokesperson told MarketWatch. Amazon generates approximately 10% of its revenue from advertising, a business that agentic shopping agents directly threaten by intercepting the product discovery and comparison stages that advertising currently monetizes. The Amazon and Shopify responses to Muse represent two distinct strategic positions on the same underlying commercial question: does an AI agent that completes purchases on a user's behalf make an e-commerce platform more or less valuable? Amazon's position is that it is a threat, specifically to the advertising revenue that depends on consumers discovering products through Amazon's own search and recommendation surfaces. Pivotal Research analyst Jeff Wlodarczak described Amazon's block as "one of the first clear signs that autonomous AI agents have the potential to significantly disrupt shopping-related advertising." He added that once users adopt a capable personal agent across email, calendar, research, and shopping, switching costs rise and the personal agent becomes the default interface, which is exactly the scenario Amazon's advertising business cannot afford. Shopify's position is the opposite. Deutsche Bank analyst Bhavin Shah said the partnership with Muse helps refute the recent bear thesis that AI agents could disintermediate Shopify and its large payments business, arguing that AI agents make it easier to find and buy products, with Shopify's Shop Pay and merchant catalog becoming infrastructure that agents depend on rather than competition they route around. "We want to give

## Newell Brands Has a New Design Center in Atlanta. AI Is Already Running Across Its Enterprise.

Date: 2026-09-22T07:42:54.650Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/newell-brands-has-a-new-design-center-in-atlanta-ai-is-already-running-across-its-enterprise

Newell Brands opened a nearly 37,000-square-foot Design Center in Atlanta on September 15, bringing AI-enabled capabilities, digital visualization, physical prototyping, and dedicated consumer research environments into one facility built to sit at the center of the company's product and brand development process. According to the company, the Design Center spans two locations. The primary facility at the Queen Tower in Concourse Office Park sits at Newell's Atlanta headquarters, co-located with brand marketing, consumer insights, innovation, and R&D teams. A photo and creative studio at the Morgan Falls location serves the content and creative production side of the operation. Together the two facilities create an integrated environment designed to compress the distance between consumer insight, design decision, and commercial execution. The AI capabilities built into the Design Center connect to a broader enterprise AI program already in operation. Newell disclosed in 2024 that it had launched AI-enabled capabilities across more than 100 enterprise use cases, applying a test-and-learn approach that validated use cases before scaling. The Design Center applies that AI foundation to the front end of product development, exploration, visualization, and rapid iteration, rather than limiting AI to operational or administrative workflows, according to the company. "Design has the power to shape how consumers experience our brands and products," said Nick Hammitt, Chief Marketing Officer at Newell Brands. "By bringing together design, consumer insights, brand teams, R&D and innovation partners, we are creating stronger connections between our consumers and brands and how we bring products to market." The Kitchen UX Lab replicates a real-world kitchen, providing a controlled but realistic environment for observing how consumers interact with cooking and food storage products across Newell's portfolio. The Kids UX Lab creates a comfortable research environment for observing parents, caregivers, and children interacting with products, covering the family-oriented brands in Newell's portfolio including Graco. Both labs give cross-functional teams direct access to consumer behavior data as part of the design and prototyping cycle rather than as a separate research phase. The intent is to surface consumer insights earlier and more frequently, before significant investment has been committed to a product direction, rather than validating designs through consumer research at a late stage when changes are expensive. "The Design Center is about bringing together talent and collaboration in one place and building the next chapter of design at Newell Brands," said Brian Rice, VP and Global Head of Design at Newell Brands. "Our

## McGraw Hill Acquired TeachFX to Put an AI Coach in Every Classroom

Date: 2026-09-22T05:41:01.448Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/mcgraw-hill-acquired-teachfx-to-put-an-ai-coach-in-every-classroom

McGraw Hill acquired TeachFX on September 17, bringing an AI-native classroom coaching tool into its K-12 portfolio and entering the professional learning market it estimates at $6 billion. According to the company, TeachFX works by analyzing classroom audio. A teacher records a session using a phone or laptop and TeachFX's proprietary AI and diarization technology, which separates and identifies different speakers within the audio, automatically processes the recording and generates feedback tied to specific instructional practices. The feedback covers high-leverage behaviors including questioning techniques, wait time, and the degree to which students are being pressed to explain their thinking. The process produces coaching insights without requiring a human coach to observe the classroom or a teacher to complete additional documentation. For school and district leaders, TeachFX surfaces aggregated classroom insights across multiple teachers, identifying instructional trends , flagging areas where curriculum implementation may be inconsistent, and enabling targeted coaching conversations rather than generic professional development. "TeachFX gives McGraw Hill a powerful new way to support educators and help them become their best professionally," said Jana Thompson, Interim President of McGraw Hill's K-12 segment. "By adding scalable instructional coaching to our portfolio, we can help educators strengthen their practice, give district leaders more actionable insights, and create a stronger bridge between instructional materials and classroom impact." TeachFX classroom feedback, the analysis generated from a recorded lesson, flows through the Model Context Protocol into Claude for Teachers, which converts it into actionable lesson plans. For educators with access to both tools, the workflow connects audio analysis directly to instructional planning without requiring manual transfer of feedback between applications. According to the press release, multiple independent studies also found that average academic performance improved for students of educators using TeachFX in math, ELA, and science. Teachers using the platform reported higher morale and a stronger sense of professional self-efficacy. "TeachFX was built to help educators better understand classroom instruction and use that insight to strengthen how they interact with students, encouraging more student talk and increasing engagement and, ultimately, outcomes," said Jamie Poskin, Co-Founder and CEO of TeachFX. "Joining McGraw Hill creates an opportunity to bring that support to more teachers, schools and districts." The acquisition completes a two-part AI strategy McGraw Hill has been building across 2026. Teachally, acquired earlier in the year, addresses the curriculum development and adaptation layer. TeachFX addresses the classroom implementation and coaching layer. Together the two acquisitions

## Nokia and Microsoft Built a Data Foundation for AI Network Agents

Date: 2026-09-22T04:39:18.151Z
Author: Sachin Mohan
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/nokia-and-microsoft-built-a-data-foundation-for-ai-network-agents

On September 17, Nokia and Microsoft announced a data platform designed to solve the problem of getting network data from multiple domains and vendors into a form that AI agents can reliably use, a process that can take weeks using traditional integration methods. The platform combines Nokia Data Suite's prebuilt telco data products with Microsoft Fabric's unified analytics and AI infrastructure, according to the companies. Nokia Data Suite provides what Nokia describes as glass-box data quality controls, making the data used by AI systems transparent and explainable rather than opaque, alongside telco-specific semantic modeling that allows individual data products to be provisioned on demand and reused across different applications. Microsoft Fabric contributes OneLake storage, Power BI, Microsoft Foundry, and Microsoft 365 Copilot. The combination is intended to give AI agents access to trusted, unified data in minutes rather than weeks, addressing what Nokia and Microsoft describe as a foundational bottleneck in moving telecom AI from experimentation to production. "Telecom providers are ready to move AI from experimentation into everyday network operations, but that requires trusted data, strong governance and platforms that can scale," said Silvia Candiani, Corporate Vice President, Worldwide Telco and Media at Microsoft. The three initial use cases named at launch include Autonomous VoNR assurance which deploys AI agents to identify voice-over-new-radio service anomalies, run root cause analysis, and recommend remedial actions using combined network, service, and subscriber observability, a 360-degree view that traditional siloed monitoring tools cannot provide. Geo-experience correlates subscriber, network, and radio-frequency data to pinpoint which users are experiencing degraded RAN performance and identify the specific coverage or capacity hotspots causing it, enabling targeted interventions rather than broad network adjustments. Predictive maintenance uses agents to analyze historical and real-time network data to surface potential issues before they reach customers, according to the companies. AI assistants operate as copilots for network engineers, providing contextual insights, recommended actions, and automated workflow execution, within clearly defined governance boundaries. Human oversight is preserved by design rather than treated as a compliance afterthought. The solution supports multi-vendor, cross-domain environments and hybrid, cloud, and on-premises deployments, allowing operators to meet regulatory requirements while advancing AI automation strategies. "Nokia and Microsoft are bringing the right data, at the right time, for the right reason to telecom operators worldwide," said Vivek Jaiswal, Senior Vice President, Autonomous Networks at Nokia. "Together, we are helping networks evolve from static infrastructures into programmable, AI-native platforms." The

## Boston Dynamics Opens Georgia Facility to Train Atlas Humanoid Robots

Date: 2026-09-21T18:57:54.467Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/boston-dynamics-opens-georgia-facility-to-train-atlas-humanoid-robots

Boston Dynamics has opened a facility inside Hyundai Motor Group Metaplant America (HMGMA) in Georgia to train its Atlas humanoid robots for automotive manufacturing, bringing the robots closer to commercial deployment on factory floors. The Robotics Metaplant Application Center (RMAC), located outside Savannah, will serve as a testbed for Atlas in real manufacturing environments. The robots are initially being trained to handle logistics and sequence automotive parts before they are delivered for assembly, Boston Dynamics said. The facility is part of Hyundai Motor Group's wider strategy to develop, train and deploy physical AI systems across its manufacturing operations. Hyundai has previously outlined a phased plan for Atlas, beginning with parts sequencing before expanding into more complex manufacturing tasks. Hyundai plans to begin full-scale Atlas deployment at HMGMA in 2028, according to its latest robotics strategy. The company expects the robots to expand to Kia AutoLand Georgia in 2029 and eventually other production facilities. From Training Center to Production Lines The RMAC gives Boston Dynamics a controlled environment for training Atlas before the robots move into production operations. Hyundai has described the facility as a place to simulate real operating conditions, collect real-world data and conduct testing and verification before robots reach production lines. The company has also been building the production infrastructure needed to scale its robotics business. Hyundai announced in January that it was targeting a production system capable of manufacturing 30,000 robots annually by 2028. Its strategy calls for Atlas to begin with parts sequencing in 2028 and expand to component assembly by 2030. That manufacturing strategy has been developing alongside Hyundai's plans for Atlas production and deployment. Boston Dynamics unveiled the production version of Atlas at CES 2026 and began manufacturing the robot at its Boston headquarters. The company said all Atlas deployments for 2026 had already been committed, with robots scheduled for Hyundai's RMAC and Google DeepMind. The company has also partnered with Google DeepMind to integrate Gemini Robotics foundation models with Atlas. The partnership is focused on enabling humanoid robots to perform a wider range of industrial tasks, beginning with automotive applications. Automakers Expand Humanoid Robot Programs Hyundai's plans place Boston Dynamics in a growing group of automakers experimenting with humanoid robots inside manufacturing operations. Toyota has outlined its own plans to use humanoid robots across its factories, while BMW has expanded its deployment of Figure AI's humanoid robots at its U.S. manufacturing operations. For

## ABB Launches Infinitus DC Portfolio for AI Data Centers

Date: 2026-09-21T10:57:26.777Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/abb-launches-infinitus-dc-portfolio-for-ai-data-centers

ABB has launched Infinitus, an integrated portfolio of direct current (DC) power solutions designed for AI data centers, as rising compute requirements push electricity demand higher. The portfolio is designed to improve energy efficiency, reduce the footprint of power infrastructure and increase the computing capacity that data centers can support. ABB expects 25% to 40% of new data-center capacity installed in 2030 could use DC distribution. The launch comes as global data-center electricity consumption is expected to rise sharply. The International Energy Agency (IEA) projects consumption to reach about 950 terawatt-hours (TWh) by 2030, roughly double the 485 TWh recorded in 2025. AI-focused data-center electricity consumption is expected to triple over the same period. The growth is already creating pressure on electricity infrastructure. New US data centers, for example, are facing constraints around grid capacity and interconnections, adding another consideration for operators planning large AI facilities. AI Racks Are Driving a Shift in Power Architecture ABB said next-generation AI chips are expected to require 1 megawatt (MW) or more per rack, compared with almost 200 kilowatts today. At those power levels, the company is positioning 800-volt direct current (800 VDC) distribution as a way to reduce conversion stages and associated heat losses. The shift is also being driven by the increasing density of AI systems. New large-scale facilities are being designed around substantially higher compute and power requirements, including NVIDIA-based systems deployed in dedicated AI campuses. ABB and Boston Consulting Group (BCG) estimated that DC distribution can deliver more than 5% in energy-efficiency gains. In a modeled 500 MW data center, ABB said a 5% efficiency improvement could make an additional 25 MW available for IT loads. The company estimated that this could support more than $300 million in additional annual revenue, although the figure is based on the report's assumptions. ABB's Infinitus portfolio is built around solid-state transformer technology and is designed as a source-to-rack DC architecture for both DC-native and hybrid AC/DC data centers. Its five main components cover medium-voltage powertrains, DC sources and power-quality systems, DC power distribution, DC power protection, and cooling optimization using DC-connected variable-speed drives and motors. ABB Builds on NVIDIA 800 VDC Work The launch extends ABB's work with NVIDIA on 800 VDC architectures for gigawatt-scale AI facilities. The companies began collaborating in 2025 on power infrastructure designed to support AI racks requiring up to 1 MW. NVIDIA has since outlined an 800

## Autodesk Expands AI Across Design and Manufacturing Workflows at AU 2026

Date: 2026-09-18T18:44:12.263Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/autodesk-expands-ai-across-design-and-manufacturing-workflows-at-au-2026

Autodesk has introduced a new set of artificial intelligence (AI) capabilities across its design and manufacturing software as it pushes deeper into automated engineering workflows. The announcements, made at Autodesk University (AU) 2026, span Fusion, Inventor, Vault and product lifecycle management (PLM), with the company focusing on repetitive design and manufacturing tasks that require engineering context. The push follows Autodesk’s strong second-quarter performance. The company reported $2.046 billion in revenue for fiscal 2027’s second quarter, up 16% year over year, and raised its full-year revenue and billings outlook. Autodesk said AI for physical-world applications requires a higher standard of accuracy than general-purpose AI because engineering decisions depend on geometry, product data, manufacturing processes, operating conditions and lifecycle requirements. Autodesk Adds AI to Design and Manufacturing Tasks In Fusion, Autodesk introduced AutoTimeline, which uses AI to convert imported geometry without design history into editable parametric parts. It also introduced AutoAssemble, which can identify parts and bring them together into assemblies. The company also introduced System Modeler, which extends automation into manufacturing by helping teams automate machine setup, toolpath generation and production preparation. Autodesk said manufacturers can create reusable automations rather than repeat the same preparation steps for every job. Autodesk is also developing an agentic PLM experience within Fusion. The system is designed to connect product data and requirements, help teams find and reuse existing work, coordinate activities and keep product information current as projects change. The approach reflects Autodesk’s wider move toward connected data and workflow automation. The company has previously described a progression from automating individual tasks to automating workflows and eventually connecting entire systems across manufacturing operations. Autodesk is also extending Autodesk Assistant into Inventor and Vault. In Inventor, Assistant can use the software’s application programming interfaces (APIs) and iLogic foundation to analyze models, modify parameters and components, create model states and automate repetitive tasks. Vault can use natural-language search to surface relevant existing parts and engineering information. AI Moves Toward Industrial Workflow Execution Autodesk’s announcements come as engineering software companies experiment with AI systems that can execute increasingly complex workflows. NVIDIA’s NemoClaw, for example, has been adopted by engineering software companies including Cadence, Dassault Systèmes, Siemens and Synopsys for autonomous engineering workflows spanning areas such as simulation, verification and design. The distinction is becoming important in manufacturing, where AI systems have to operate against physical constraints. Caterpillar has also combined engineering and operational data with AI-powered robotics

## SK Hynix in Talks With Intel to Make Memory Chips in the US for First Time

Date: 2026-09-18T18:00:06.894Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/sk-hynix-in-talks-with-intel-to-make-memory-chips-in-the-us-for-first-time

SK Hynix is in exploratory talks with Intel about manufacturing memory chips in the United States for the first time, potentially using part of Intel’s planned Ohio chipmaking facility, according to three people familiar with the discussions cited by Reuters . One option under discussion would have SK Hynix lease part of Intel’s Ohio facility. Another could involve a venture between the two companies and major cloud providers seeking to secure memory supplies. The talks remain exploratory, and no agreement has been finalized. SK Hynix confirmed that it was reviewing options to strengthen the competitiveness of its memory business but said no specific plans or arrangements had been decided. In a separate statement, the company said nothing had been finalized regarding cooperation with companies mentioned in the report or memory chip production in the U.S. The potential deal comes as AI data center construction increases demand for memory. AI data centers are eating away at the world's memory chips, with manufacturers shifting capacity toward high-bandwidth memory (HBM) and other products used in AI systems. Why SK Hynix Is Looking at U.S. Production SK Hynix produces dynamic random-access memory (DRAM) for servers, PCs and smartphones, along with NAND flash memory used for storage. It is also a major supplier of HBM, which is used alongside AI processors. The company already has a U.S. manufacturing presence under development. In August, SK Hynix broke ground on an advanced packaging facility in Indiana, with an investment of more than $4 billion. The facility is scheduled to begin mass production of next-generation HBM in the second half of 2029. Wafers produced in South Korea will be sent to Indiana for advanced packaging and testing. Fabricating memory chips in the U.S. would therefore represent a further expansion of SK Hynix’s American manufacturing footprint. The company faces higher labor and construction costs in the U.S., while much of the semiconductor supply chain remains concentrated in Asia, according to the Reuters sources. Still, customer and government pressure is pushing the company to consider additional U.S. capacity. SK Group Chairman Chey Tae-won said in July that SK Hynix faced significant pressure from customers and governments to supply more chips and that he believed the company should build a factory in the U.S., if possible. Intel’s Delayed Ohio Project Could Provide Capacity For Intel, a potential arrangement could give greater use to infrastructure at its Ohio project, which has faced

## Enterprises Must Put Business Problems Ahead of AI Hype

Date: 2026-09-18T11:06:57.830Z
Author: Deployed
Desk: Retail
URL: https://aimmediahouse.com/retail/enterprises-must-put-business-problems-ahead-of-ai-hype

Enterprises are not short on AI ambition, but many still struggle to build the talent, processes, governance and operating models needed to turn experiments into lasting business value, according to industry leaders. Enterprises are pushing to scale AI, but many are finding that ambition alone does not translate into business value. Leaders from Tata Play, PayPal, Western Governors University and the data science industry point to gaps in talent, processes, governance, data and operating models as key barriers to scaling AI effectively. Across their views, one message stands out. Scaling AI requires more than deploying new models or increasing the number of pilots. Organisations need to rethink workflows, focus on validated business problems and build the foundations required to move AI from experimentation into production. AI Readiness is a Layered Problem Krupa Shah, Head of Analytics at Tata Play, says AI readiness in large enterprises involves multiple areas that need attention, but identified three as particularly important: experienced talent, execution speed and responsible AI governance. “Availability of AI-experienced talent who also understands the business/product” remains a key challenge, Shah says. Enterprises need people who can connect AI capabilities with specific business and product requirements rather than treating AI as a standalone technology function. She also highlights the need to move faster in experimentation and deployment. “Fail fast, experiment multiple POCs and speed to production” should be part of how enterprises approach AI, she says. At the same time, faster deployment needs to be accompanied by “governance around responsible” AI, making governance a parallel requirement rather than something addressed after deployment. Start With the Problem, Not the Technology For Bibhash Chakrabarty, Senior Director, AI and Decisioning at PayPal, the biggest mistake is more fundamental. Enterprises often start building solutions before they have properly understood the problem they are trying to solve. “The biggest misstep is rushing to solve before the problem is properly understood,” Chakrabarty says. He notes that many enterprise problems are not purely technical. They are often embedded in processes that have become outdated or inefficient. Applying GenAI or agentic AI to those processes without redesigning them can produce an impressive demonstration but fail to deliver when the solution needs to scale. “Dropping a GenAI or agentic AI solution on top of that process without redesigning it leads to a cool demo or pilot but struggles with scaling,” he says. For Chakrabarty, closing the gap between AI ambition and

## Toyota Plans 400,000 Humanoid Robots for Factory Automation

Date: 2026-09-18T06:43:37.519Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/toyota-plans-400000-humanoid-robots-for-factory-automation

Toyota Motor Corp. is making a full-scale entry into the AI robotics business, starting with humanoid-style robots designed to work alongside factory employees. The automaker recently demonstrated one of its robots at its European research and development headquarters in Brussels. Toyota plans to introduce the systems gradually at its own production facilities and those of major business partners as factories undergo renewal, according to The Yomiuri Shimbun. The Yomiuri Shimbun report Nikkei Asia reported that Toyota plans to deploy about 400,000 humanoid robots to work alongside factory staff. The scale would make the initiative one of the larger planned deployments of humanoid robots in industrial settings. Toyota has been developing the technology around what it calls physical AI, which allows robots to perceive their surroundings through sensors, learn from human movements and adjust their actions to physical conditions. Its latest research robot , ELEY, was designed with human-like joint structures to make it easier to reproduce human movements. Toyota also added a scapular axis to improve the robot's range of motion during tasks involving two-handed movements. The approach resembles the growing use of humanoid robots by other automakers. BMW, for example, has expanded its deployment of Figure AI's humanoid robots at its Spartanburg plant, moving from body-shop applications into logistics work on the assembly floor. Toyota Wants Robots to Assist Workers Toyota's robotics strategy is tied to factory automation and labor shortages, while the company has emphasized using robots as tools that work alongside people. At a technical briefing held by Toyota Motor Europe earlier this month, Toyota Executive Vice President Hiroki Nakajima said the company wanted to expand factory automation as a business and contribute through manufacturing. Toyota has been developing the underlying technology for years. Its Human Support Robot (HSR), first unveiled in 2012, has provided a research platform for autonomous navigation, object recognition and motion execution. Some of those technologies have already been applied to KumiPro, a factory robot used at Toyota Motor East Japan. The company is also working on robotics foundation models that learn from robot movement and sensor data collected in physical environments. Toyota said it had collected about 350 hours of data from eight locations as of March 2025, while noting that insufficient data remained a constraint on making the models flexible across different environments. Toyota formally included robotics among the initial focus areas of its Global AI Accelerator, alongside manufacturing, autonomous driving,

## Four AI Labs Shipped New Models in One Week. Enterprise Buyers Are Done Keeping Up.

Date: 2026-09-18T04:28:05.709Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/four-ai-labs-shipped-new-models-in-one-week-enterprise-buyers-are-done-keeping-up

Four of the world's largest AI labs shipped new frontier models in the same week. That is not a coincidence, it is a commercial strategy. And for the enterprise technology leaders responsible for evaluating, procuring, and integrating those models, it has a specific name: model fatigue. In the first week of September 2026, Anthropic released Claude Fable 5.1 and Mythos 5.1. Meta released Muse Spark 1.3. Google released Gemini 3.8 Flash. OpenAI released GPT-6 Astra. Each release came with benchmark improvements, new capability claims, and updated pricing. Each release triggered a new evaluation cycle at every enterprise currently running an AI program. OpenAI CEO Sam Altman told CNBC directly: "We're all moving to faster cadences." "I feel like model fatigue is a real thing," said Zhen Lu, CEO of Runpod. "It's really challenging to go evaluate every one of the ones that are coming out right now," said Suresh Vasudevan, CEO of Clockwork Systems. "If my startup wants to evaluate 10 AI models for a certain task, it may just pick five." For a startup, picking five instead of ten is an acceptable trade-off. For a bank, an insurer, or a healthcare system running production AI workflows, it is not that simple. The Procurement Math Nobody Is Publishing Enterprise AI evaluation does not move at lab speed. Security review, compliance assessment, integration testing, and internal approval, the full enterprise procurement cycle for a new AI model takes four to six months. Labs are shipping on an eight to twelve week cadence. The arithmetic produces a structural problem. By the time an enterprise finishes evaluating Model A, Models B and C have already shipped. The enterprise is perpetually one or two generations behind the current frontier. The cost of that mismatch does not appear on any vendor's pricing page. It lands in engineering organizations, the teams that have to re-run evaluations every time a model version changes, rebuild integrations when APIs shift, and restart compliance processes when model behavior updates alter outputs that were previously approved. A 2025 Menlo Ventures analysis found Anthropic's share of enterprise LLM API spend went from 12% in 2023 to 40% in 2025, while OpenAI's fell from 50% to 27%. Whatever model an enterprise standardized on in 2023 was not the market's strongest choice by 2025. There is no reason to expect 2027 will look like 2026. The lab incentive structure runs in the opposite direction

## Sentara Health Pilots GW RhythmX AI Platform for Primary Care

Date: 2026-09-17T18:51:47.749Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/sentara-health-pilots-gw-rhythmx-ai-platform-for-primary-care

Sentara Health is piloting an artificial intelligence platform from GW RhythmX to provide patient-specific clinical recommendations during primary care visits, expanding the health system’s use of AI from administrative tasks into point-of-care decision support. The 90-day pilot will involve 30 physicians and advanced practice providers (APPs) in family and internal medicine across two markets. Sentara will evaluate clinician adoption and trust, the usefulness of the recommendations and how well the technology fits into existing workflows before considering a wider rollout. Sentara operates 12 hospitals and more than 400 care sites across Virginia and Northeast North Carolina. The not-for-profit health system reported $14.2 billion in operating revenue in 2025 and has approximately 473 primary care physicians and APPs within Sentara Medical Group. AI Moves Into Primary Care Workflows GW RhythmX was formed in 2025 through the combination of RhythmX AI and Get Well. Its precision care platform uses generative and predictive AI to analyze information including patient history, clinical guidelines, care gaps, social and financial determinants, payer requirements and formulary considerations. The system then presents patient-specific recommendations to clinicians during an encounter. GW RhythmX describes these as next-best-action recommendations intended to help care teams identify opportunities for intervention and coordinate care. The approach reflects a growing effort by health systems to put clinical AI directly into existing workflows. Sutter Health has also moved AI into its EHR workflows, integrating OpenEvidence into Epic so physicians can access clinical evidence while working inside the electronic health record. Sentara’s pilot is focused on the cognitive workload associated with primary care. Prasanna Mohanty, Executive Vice President and President for Ambulatory Care at Sentara Health, said the health system was redesigning its ambulatory model around team-based care involving physicians, APPs, nurses, pharmacists and care managers. “We're looking at RhythmX AI to augment that model by reducing not only administrative burden, but the cognitive burden,” Mohanty said. The system will also run a control group using existing Sentara tools. A separate pilot of Sentara’s team-based primary care model will run alongside the AI deployment, allowing the health system to assess the effects of the two initiatives independently before combining them as they scale. Sentara Will Test Clinical Impact Before Scaling The pilot comes as healthcare organizations expand AI beyond documentation. AI scribes are moving closer to routine healthcare use , with ambient tools primarily focused on reducing the time clinicians spend preparing medical notes. GW RhythmX is

## OpenEvidence and Penn Medicine Partner to Bring AI Clinical Decision Support to Botswana

Date: 2026-09-17T12:23:06.638Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/openevidence-and-penn-medicine-partner-to-bring-ai-clinical-decision-support-to-botswana

Penn Medicine is partnering with medical AI company OpenEvidence to deploy AI-powered clinical decision support to more than 10,000 clinicians across its seven-hospital U.S. health system and extend the technology to clinicians in Botswana. The partnership will provide clinicians with personalized medical information drawn from peer-reviewed literature and medical societies, with OpenEvidence tools accessible within Penn Medicine and the Botswana-UPenn Partnership (BUP). The system is designed to surface evidence alongside information from the electronic health record (EHR) within existing clinical workflows. OpenEvidence has also worked with health systems to embed clinical AI directly into EHR workflows. Sutter Health's integration of OpenEvidence with Epic is one example of the technology being used to provide physicians with access to clinical information during care delivery. The Penn partnership expands that approach across the health system while introducing a separate development effort for a lower-resource clinical setting. Extending Clinical AI to Botswana The international component will operate through BUP, a 25-year-old academic health partnership between Penn, the University of Botswana and the Botswana Ministry of Health. The program began in 2001 and has supported clinical training, postgraduate medical education and research in areas including HIV/AIDS, oncology and population health. Under the agreement, Penn Medicine teams and BUP clinicians will work on a customized set of OpenEvidence tools designed around clinical conditions and constraints in Botswana. “Building a customized set of tools in direct partnership with BUP clinicians and partners reflects a shared conviction that effective clinical AI will be shaped by local expertise and adapted for the conditions they practice in,” Farouk Dako, MD, MPH, Director of the Botswana-UPenn Partnership and Strategic Advisor for Academic Partnerships at the University of Pennsylvania Perelman School of Medicine Center for Global Health, said. The partnership comes as Penn Medicine expands its use of AI across clinical operations. The health system previously deployed K Health's clinical agents across its virtual care network, adding to its use of AI in patient-facing and clinical workflows. Penn Medicine also operates Penn Pathways, electronic consultation programs and brick-and-mortar and virtual care programs that could incorporate AI-based clinical decision support, according to the announcement. OpenEvidence Expands Its Clinical Footprint OpenEvidence's partnership with Penn adds to its work with medical organizations and clinicians. The company has also partnered with the Anticoagulation Forum on AI-guided clinical decision support for antithrombotic care, extending its use into specialized areas of medicine. For Penn clinicians, the new deployment

## Nokia's AI-RAN Platform Is in Live Trials Across Four Continents

Date: 2026-09-17T11:36:20.242Z
Author: Sachin Mohan
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/nokias-ai-ran-platform-is-in-live-trials-across-four-continents

Nokia announced on September 16 that eight operators across four regions are running AI-RAN trials on its platform, extending a footprint that already includes T-Mobile, SoftBank, Indosat Ooredoo Hutchison, and NTT DOCOMO into new markets spanning the Middle East, Asia-Pacific, Europe, and North America simultaneously. Nokia's AI-RAN platform has already delivered more than 20% improvements in spectral efficiency, the measure of how much data a network can transmit over a given frequency band, through AI-powered radio innovations. A software roadmap extending through 2027 and 2028 targets further gains from the same infrastructure without requiring hardware replacements, according to the company. For operators whose capital expenditure cycles are built around hardware refresh timelines, the ability to extract more capacity from existing spectrum through software upgrades has a direct and quantifiable impact on network economics. "AI is reshaping the future of telecommunications and creating a significant opportunity for operators to rethink the role of the radio network," said Emma Falck, President of Mobile Infrastructure at Nokia. "The increasing engagement we are seeing from service providers around the world shows that AI-RAN is moving quickly from vision to commercial reality and becoming a foundational technology for the AI-native 6G era." According to the company, the technical foundation of Nokia's AI-RAN platform is the combination of its AI-native anyRAN software with NVIDIA Aerial RAN Computer, a purpose-built GPU-accelerated platform for radio access network workloads. The architecture is designed to handle both connectivity and AI workloads on the same infrastructure, enabling operators to run AI applications at the network edge alongside standard telecommunications traffic rather than maintaining separate infrastructure for each. Operators deploying Nokia's AI-RAN platform for 5G Advanced today are buying infrastructure that can be upgraded to AI-native 6G through software rather than hardware replacement, compressing the capital investment required for the next network generation into a software cost rather than a full infrastructure refresh. "AI-RAN will transform the mobile network into one of the world's largest distributed AI infrastructures," said Ronnie Vasishta, Senior Vice President of Telecom at NVIDIA. "With NVIDIA Aerial RAN Computer and Nokia's AI-RAN software, operators can build software-defined networks that continuously improve in performance and efficiency, while supporting entirely new AI services at the edge." The operator list now spans twelve named partners across two generations of trial stage, eight newly announced alongside four previously disclosed, covering markets that represent a significant share of global mobile network investment.

## Groupon Is Opening Its Platform to AI Agents and Third-Party Builders

Date: 2026-09-17T02:06:18.615Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/groupon-is-opening-its-platform-to-ai-agents-and-third-party-builders

Groupon announced on September 15 that it is opening its platform to external builders and AI agents, converting its deal inventory and checkout infrastructure into a foundation that third parties can build on rather than a closed product only accessible through Groupon's own surfaces. The announcement came via CEO Dušan Šenkypl on LinkedIn , where he outlined the commercial logic directly. Groupon already has live integrations in ChatGPT and Claude. The AI Builder Pack, expected to go to public access in October, takes that further by allowing any developer, student, or community organizer to build a product using Groupon deals and earn commission on the purchases that flow through it. Groupon handles the inventory and checkout. The builder handles the audience and the surface, according to the company. "Groupon should be accessible wherever customers choose to spend their time: websites, apps, AI assistants and personal agents," Šenkypl wrote. "A student, an independent developer or a community organiser should be able to turn an idea into something useful." The four use cases Šenkypl named are deliberately varied in scale, from an individual building a neighborhood deals website to a community organization funding a charity through earned commissions. By opening its deal inventory and checkout infrastructure to external builders, Groupon says it is turning its supply layer into a distribution platform. Every website, app, or AI agent built on top of it becomes a new channel through which Groupon deals reach customers it would not otherwise have reached. Šenkypl explicitly named Meta's Muse, the 24/7 personal AI agent launched earlier this month, as evidence of where the market is heading. As AI agents become an increasingly common way consumers discover and purchase local experiences, a deals platform that is only accessible through its own app and website will miss a growing share of AI-mediated transactions. The ChatGPT and Claude integrations are Groupon's first step. The Builder Pack is the infrastructure for the next one, allowing AI agents built by anyone to connect to Groupon's inventory rather than waiting for Groupon to build each integration itself. The model keeps Groupon's cost structure light, builders bear the development and distribution cost, Groupon bears the deal sourcing and checkout infrastructure cost, and commission aligns both parties around transaction volume. Groupon's version is scoped to local deals and community use cases, with a commission model that aligns builder incentives with transaction volume rather than platform subscriptions.

## Novo Nordisk Turns to Anthropic to Accelerate AI-Driven Drug Development

Date: 2026-09-16T18:46:28.020Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/novo-nordisk-turns-to-anthropic-to-accelerate-ai-driven-drug-development

Novo Nordisk has entered a collaboration with Anthropic to accelerate drug discovery and development, adding Claude models to the Danish drugmaker’s growing artificial intelligence strategy. Under the agreement announced September 16, Novo Nordisk and Anthropic will identify challenges in drug discovery, develop solutions for specific scientific workflows and use AI to support biological reasoning. Financial terms were not disclosed. The companies will initially test Claude Science on selected research and development workflows where they expect the technology to have the greatest impact. Anthropic introduced Claude Science in June as an AI workbench that connects researchers with scientific databases, computational tools and research workflows. The platform supports areas including genomics, proteomics, structural biology and cheminformatics. Novo Nordisk CEO Mike Doustdar said the partnership would “supercharge our R&D organisation” as the company seeks to expand AI across its research operations. The company also said Anthropic’s models will support software development, allowing it to increase the use of AI across the organization. Novo Builds a Multi-Model AI Strategy The Anthropic collaboration follows several other AI initiatives at Novo Nordisk. In April, the company entered a strategic partnership with OpenAI to deploy AI across drug discovery, manufacturing, supply chain, distribution and commercial operations. Novo Nordisk also established a strategic partnership with Amazon Web Services (AWS) in August. The agreement includes a co-innovation hub in London where AWS engineers and scientists are working with Novo Nordisk research teams on AI technologies intended to accelerate drug discovery. AWS became Novo Nordisk’s preferred cloud provider and strategic AI partner under the agreement. The company has described these initiatives as part of its goal to become “the world’s most AI-driven healthcare company.” The Anthropic partnership adds another frontier AI provider to that strategy, with the companies focusing specifically on scientific research as well as software engineering. Novo Nordisk has already used Claude technology in internal workflows. Anthropic’s case study on the company describes NovoScribe, a generative AI platform developed using Claude Code that supports the production of clinical study documentation. Anthropic reported that the system reduced the time required for some clinical study documentation from more than 10 weeks to 10 minutes. Those figures are based on Anthropic’s account of Novo Nordisk’s deployment. Pharma Pushes AI Deeper Into R&D Novo Nordisk’s move comes as pharmaceutical companies expand AI from individual research applications into larger scientific and computing environments. Bristol Myers Squibb, for example, has deployed NVIDIA infrastructure to

## Rockwell Automation Joins Anthropic’s Project Glasswing to Advance AI Cyber Defense

Date: 2026-09-16T18:01:00.000Z
Author: Deployed
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/rockwell-automation-joins-anthropics-project-glasswing-to-advance-ai-cyber-defense

Rockwell Automation has joined Anthropic’s Project Glasswing, a cybersecurity initiative that gives approved organizations access to advanced AI models for defensive security work. The industrial automation company announced the partnership on 09/15. Through controlled access to Claude Mythos 5, Rockwell’s security teams are exploring ways to accelerate the discovery, validation, prioritization, and remediation of vulnerabilities across software and connected systems used by manufacturers. The initiative puts an AI model developed for cybersecurity into the security workflow of a company whose products are used across industrial automation environments. Rockwell said the work is focused on identifying and addressing vulnerabilities while maintaining a controlled, defensive approach. “Rockwell Automation plays a central role in helping manufacturers operate safely, securely, and reliably,” said Tony Baker, Vice President and Chief Product Security Officer, Digital Trust, Rockwell Automation. “By participating in Project Glasswing, we are applying advanced AI capabilities in a controlled, defensive way to help identify and address vulnerabilities faster and support the resilience of the industrial ecosystem.” Anthropic Expands Project Glasswing Anthropic launched Project Glasswing in April 2026 to provide approved organizations with access to Claude Mythos Preview for defensive cybersecurity work. The company initially brought roughly 50 partners into the initiative and later expanded it to approximately 150 additional organizations across more than 15 countries. The expansion included organizations supporting critical infrastructure in areas such as power, water, healthcare, communications, and hardware. Anthropic has described the program as an effort to help defenders find vulnerabilities earlier and address them before attackers can exploit them. The initiative builds on Mythos’s ability to perform cybersecurity tasks that traditionally require significant amounts of specialized expertise and manual analysis. Anthropic reported that participants in the initial phase had identified thousands of high- and critical-severity vulnerabilities. AIM Media House previously examined the technology behind the program in its report on Anthropic’s AI model exploiting software vulnerabilities at scale, providing context for the capabilities now being applied in industrial environments. Anthropic has continued to restrict access to its most capable cybersecurity models to vetted organizations because of the potential risks associated with advanced automated vulnerability discovery and exploitation. AI Moves Into Industrial Cybersecurity Rockwell’s participation extends the use of Project Glasswing into industrial automation, where vulnerabilities can affect connected software, control systems, production environments, and the operations that depend on them. Rockwell’s 2025 annual report identified cybersecurity as a material consideration for the company because its products are used

## Cadence and Analog Devices Develop DSP Architecture With 5X Performance Boost

Date: 2026-09-16T17:31:16.380Z
Author: Deployed
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/cadence-and-analog-devices-develop-dsp-architecture-with-5x-performance-boost

Cadence and Analog Devices (ADI) have jointly developed a new digital signal processing (DSP) architecture for ADI’s next-generation automotive audio processors, with the companies targeting higher performance for AI-enabled audio, voice and signal-processing workloads. Cadence announced the collaboration on 09/15, saying its Tensilica DSP technology has been integrated into ADI’s new ADSP-SC84x/2184x platform. The new SHARC-FX processor core delivers 5X higher performance than the SHARC+ processor used in the previous-generation ADSP-SC59x family, according to Cadence. The platform is designed for automotive applications including road-noise and engine-order cancellation, personal sound zones, immersive 3D audio, voice processing and hands-free communications. ADI is also targeting applications such as intelligent voice assistants, AI-enhanced noise cancellation and generative AI audio. New DSP Architecture Targets AI Workloads The ADSP-SC84x/2184x platform moves from a 28nm process to 16nm and adds an LPDDR4 memory interface, increased on-chip L2 memory and a dedicated hardware security module. ADI is also offering optional connectivity cores for enhanced communications and mainline Linux support. The SHARC-FX core was developed through a joint engagement between Cadence and ADI. Its architecture supports a wider range of data types, an enhanced very long instruction word (VLIW) single instruction, multiple data (SIMD) architecture, improved cache operations and an expanded instruction set for signal processing and AI workloads, Cadence said. The new architecture is intended to allow more processing to take place on the DSP itself. Cadence said workloads such as music genre identification, music source separation and real-time vocal removal can run on the DSP while the host CPU handles vehicle-domain and operating system functions. The move follows a wider push to run AI capabilities directly inside vehicles. Mercedes-Benz, for example, has been working to bring speech processing, language understanding and reasoning onto vehicle hardware through its collaboration with Liquid AI, rather than relying exclusively on cloud processing. Cadence said the SHARC-FX core is designed to support real-time neural-network inference at the edge alongside conventional DSP workloads. It supports optimized neural-network kernel routines for AI and machine learning frameworks including Google’s LiteRT, allowing models to be offloaded from the host CPU. Designed for Software-Defined Vehicles The ADSP-SC84x/2184x platform is aimed at modern software-defined vehicles, where audio, voice and other vehicle functions increasingly depend on software running across different computing resources. The approach also fits with the growing use of AI within vehicle operating systems. Mercedes-Benz’s MB.OS architecture, for instance, brings multiple AI models into its MBUX infotainment

## Estée Lauder Partners with Profound to Optimize Brand Visibility Across AI Platforms

Date: 2026-09-16T11:56:08.517Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/este-lauder-partners-with-profound-to-optimize-brand-visibility-across-ai-platforms

On September 14, Estée Lauder announced a global strategic partnership with Profound, an AI marketing platform, to monitor and improve how its portfolio of brands is represented across generative AI platforms as consumers increasingly turn to tools like ChatGPT and Gemini to discover products and seek beauty recommendations before a purchase. The partnership gives Estée Lauder a unified view of how its brands appear across major AI search platforms globally, including when and how they are recommended to consumers and where opportunities exist to improve visibility and accuracy, according to the company. The company will use Profound's agentic tools to optimize content across product description pages, blogs, YouTube channels, and digital assets at scale, ensuring product information, ingredients, claims, and benefits are structured in ways that are easier for large language models to surface to consumers. Estée Lauder said it is the first company in prestige beauty to deploy Profound's capabilities across its full global portfolio, operationalizing a practice the industry refers to as Generative Engine Optimization. "This partnership gives every brand in our portfolio the ability to show up distinctly, consistently, and accurately wherever our consumers are asking for recommendations," said Aude Gandon, Global Chief Digital and Marketing Officer at Estée Lauder. Generative Engine Optimization is emerging as the AI-era equivalent of search engine optimization - the discipline of ensuring brand content is structured and positioned in ways that AI answer engines can accurately retrieve and recommend. The shift matters because AI assistants do not return ranked lists of links the way traditional search engines do. They synthesize answers, and the brands that surface in those answers are determined by how well their content is understood by the underlying models. James Cadwallader, CEO and co-founder of Profound, said Estée Lauder is approaching GEO the way the best marketers approach any platform shift - early, comprehensively, and with real operational commitment. Profound is used by 16% of the Fortune 500, including Comcast and Walmart. Estée Lauder sells products across approximately 150 countries under brands including Estée Lauder, Clinique, M·A·C, La Mer, and Too Faced.

## TripGenie AI Orders Grew 400%. 60% of Its Interactions End in a Booking.

Date: 2026-09-16T11:32:58.348Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/tripgenie-ai-orders-grew-400-60-of-its-interactions-end-in-a-booking

TripGenie launched in July 2023 as a conversational travel planning and booking assistant. By Q2 2026, it had become one of Trip.com's primary booking channels. Executive Chairman James Liang disclosed on the company's September 15 earnings call that AI-assisted orders through TripGenie increased approximately 400% year over year, and that nearly 60% of TripGenie interactions are now booking-related, covering hotels, flights, and attractions. "We are seeing growing evidence that AI is becoming an integral part of the travel journey," Liang said. "Travelers are increasingly relying on AI not only for inspiration, but also to make and fulfill travel decisions." Trip.com also rolled out fully AI-powered search across its platform in Q2, designed to interpret user intent more accurately and surface more relevant results, with early indicators showing higher platform engagement following the rollout. CEO Jane Sun responded to a direct analyst question about agentic AI in travel with a four-stage framework for how Trip.com views AI's role across the full travel journey. The first stage is Inspiration, where AI agents are becoming an increasingly important discovery channel as they improve at understanding preferences and generating personalized recommendations. Trip.com is expanding partnerships with leading AI platforms and applying GEO and AEO strategies to capture AI-driven traffic as it grows, according to the company. The second stage is Search, where Trip.com's proprietary travel data becomes the competitive advantage. "This is where our proprietary travel data and supply capabilities become increasingly valuable," Sun said. "Our data covers not only hotel and flight information but also inventory, pricing, availability, policies, and other attributes needed to make and fulfill a booking." In Q2, the company also explored agent-to-agent collaboration to capture emerging AI-driven demand. The third stage is Transaction, where Trip.com's focus is a closed-loop flow connecting real-time inventory, pricing, payment, and confirmation with minimal friction. The fourth is Fulfillment, where AI handles service support and change management while human operational execution stays at the center. The underlying argument across all four stages is that AI changes the discovery interface but does not eliminate the operational complexity of delivering travel. Inventory, real-time pricing, transaction infrastructure, and on-the-ground support all remain essential regardless of how a trip is discovered. Trip.com's position is that its decade-long investment in those capabilities makes its platform more valuable in an AI era, not less. CFO Cindy Wang confirmed the AI investment model is deliberately scoped. "Most of our work is

## Only 12% of CEOs Say AI Has Delivered Both Cost and Revenue Benefits

Date: 2026-09-16T05:46:01.823Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/only-12-of-ceos-say-ai-has-delivered-both-cost-and-revenue-benefits

Most enterprises are spending more on AI than ever before. Most are not seeing the return. PwC surveyed 4,454 chief executives across 95 countries in January 2026. The question was simple: what did AI do to your revenue and your costs over the last twelve months? The biggest single answer, at 56%, was nothing. Not slower growth than expected. Not marginal improvement. Nothing. More than half of the world's CEOs reported that AI delivered neither higher revenue nor lower costs in the past year. Twelve percent reported both. PwC calls them the vanguard. And roughly 13% came out worse. Costs went up. Revenue did not follow. That last number is the one that rarely appears in coverage. More than one in ten CEOs who deployed AI ended the year with a worse cost structure, and nothing on the revenue side to show for it. What the 12% Did Differently The vanguard is a different decision. CEOs reporting both cost and revenue gains are two to three times more likely to have embedded AI extensively across products and services, demand generation, and strategic decision-making. Not in one function. Not in one pilot. Extensively, across the business. Also, 44% of vanguard companies have applied AI to their products, services, and customer experiences. Among all other companies, that figure is 17%. The gap between 44% and 17% is a deployment decision gap. The vanguard did not have better models or bigger budgets. They embedded AI deeper. PwC's separate analysis quantifies the commercial result of that decision. Companies applying AI widely to products, services, and customer experiences achieved nearly four percentage points higher profit margins than those that did not. And organizations that built strong AI foundations are three times more likely to report meaningful financial returns than those that did not. "A small group of companies are already turning AI into measurable financial returns, while many others are still struggling to move beyond pilots. That gap is starting to show up in confidence and competitiveness, and it will widen quickly for those that don't act,” said PwC Global Chairman Mohamed Kande. The failure mode behind the 56% and the 13% is the same. Deploying AI in isolated workflows without building the foundation across the business generates neither the cost efficiency that requires scale nor the revenue growth that requires AI in the customer-facing layer simultaneously. What This Week's Earnings Calls Showed Last week,

## GE HealthCare Launches AI Tool to Forecast Hospital Capacity Bottlenecks

Date: 2026-09-15T18:14:13.067Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/ge-healthcare-launches-ai-tool-to-forecast-hospital-capacity-bottlenecks

GE HealthCare has launched CareIntellect for Operations, an artificial intelligence (AI)-enabled Software-as-a-Service (SaaS) application designed to help health systems forecast capacity constraints up to 72 hours in advance. The application analyzes hundreds of patient and operational data points in real time to provide unit-, department- and enterprise-level views of emerging bottlenecks. It also recommends actions intended to help hospital teams manage capacity and patient throughput, according to GE HealthCare . The Queen's Health Systems in Hawaii and Duke Health in North Carolina will be the first health systems to evaluate the application and provide clinical and operational feedback. The launch comes as hospitals face higher patient volumes and rising operating costs. The American Hospital Association (AHA) reported that inpatient volumes increased 5.3% in 2025, while total hospital expenses increased 7.5%. Workforce costs rose 5.6% during the year, according to the AHA's 2026 Costs of Caring report . Healthcare organizations are also dealing with persistent emergency department capacity pressures. Data released by the Centers for Medicare & Medicaid Services (CMS) in August includes state-level measures of emergency department wait times and other timely-care metrics. CMS data shows that emergency department care remains a significant component of hospital capacity management. The growing focus on capacity is also reflected in healthcare AI deployments as health systems increasingly evaluate AI through operational measures such as workflow efficiency and the ability to see more patients. AI Models Forecast Pressure and Discharge Timing CareIntellect for Operations combines two proprietary GE HealthCare AI models: Pressure Forecast and Estimated Day of Discharge. Pressure Forecast uses historical and real-time operational data to predict capacity constraints across hospitals, departments and units up to 72 hours ahead. Estimated Day of Discharge forecasts when individual patients are likely to leave the hospital using longitudinal patient data. The application evaluates factors including bed availability, patient delays, staffing, wait times and ancillary services. It updates discharge estimates hourly and dynamically prioritizes recommended workflows as conditions change. The predictive approach builds on GE HealthCare's existing Command Center software, which the company says is used by almost 500 facilities worldwide. The company has developed the new application from more than two decades of experience in hospital operations and data from its Command Center platform. The discharge component of CareIntellect also fits into a broader use of AI for patient flow. Nebraska Medicine has used AI-powered capacity management to identify patients who may be ready for discharge

## Crusoe Opens Second Tulsa Manufacturing Facility to Expand AI Infrastructure Production

Date: 2026-09-15T17:57:17.856Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/crusoe-opens-second-tulsa-manufacturing-facility-to-expand-ai-infrastructure-production

Crusoe, an AI infrastructure company, has opened its second manufacturing facility in Tulsa, Oklahoma, adding 400,000 square feet to its local operations as demand for infrastructure supporting AI data centers continues to grow. The new facility brings Crusoe’s national manufacturing footprint to roughly 1 million square feet. The company manufactures medium-voltage switchgear, low-voltage switchboards, electrical enclosures, industrial controls and custom-fabricated copper busbar across its two Tulsa sites. The expansion comes as the U.S. AI data-center buildout faces increasing demands for electricity and supporting infrastructure. Power constraints are becoming a growing challenge for new AI data centers, particularly as operators seek to bring large computing facilities online more quickly. Crusoe Expands Tulsa Workforce Crusoe has hired more than 300 local craft workers across its two Tulsa facilities, compared with the 100 workers it expected when it opened its first facility in 2024. The workforce includes employees involved in engineering, welding, powder coating, fabrication and electrical assembly. The company said the Tulsa operation supports gigawatts of AI infrastructure. Its equipment has also been supplied to Crusoe’s Stargate Project site in Abilene, Texas, including more than 2,500 switchboards. The scale of the Abilene project reflects the infrastructure requirements behind AI expansion. Crusoe announced in March 2025 that its Abilene campus would expand to eight buildings, about 4 million square feet and 1.2 gigawatts of capacity. The project has also become part of the broader U.S. data-center expansion and grid capacity debate. Cully Cavness, Co-founder, President and Chief Strategy Officer of Crusoe, said the company’s manufacturing operations rely on skilled workers in Tulsa to build equipment needed for AI infrastructure. “This team creates the equipment that makes the future of AI possible, and manufacturing ourselves, here in America, is what lets us move as fast as we do,” Cavness said. Crusoe said building the equipment in-house gives it greater control over quality and production and allows it to respond more quickly to infrastructure requirements. The company said its manufacturing model can turn around equipment in days compared with timelines that can take months elsewhere in the industry. Second Facility Adds Manufacturing Capacity The new facility was unveiled during a ribbon-cutting ceremony attended by Oklahoma Gov. Kevin Stitt, Tulsa Mayor Monroe Nichols and other state and local officials. Stitt said the expansion would support more than 300 Oklahoma workers as the state seeks to expand its technology and manufacturing base. Andrew Sweatman, Vice President of

## Anthropic Launched Claude for Financial Advisors. Nine Platforms Already Connected.

Date: 2026-09-15T16:02:38.078Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/anthropic-launched-claude-for-financial-advisors-nine-platforms-already-connected

Anthropic launched Claude for Financial Advisors on September 14, connecting Claude to nine named platforms across the financial adviser technology stack and building workflow skills designed to handle the administrative work that sits between client interactions. The product connects to BlackRock, Charles Schwab, Addepar, Envestnet, iCapital, Orion, Wealthbox, Wealth.com, and Zocks, covering the custodial, portfolio analytics, CRM, and planning layers that financial advisers use daily, according to the company. The workflow skills built on top of those connections address three specific tasks: preparing for client meetings, analyzing portfolios, and running compliance checks on client-facing language before it goes out. The compliance skill is the most commercially specific element of the product. It screens client-facing language for potential regulatory issues, helps firms document their AI review activities, and helps firms document their use of AI specifically under SEC rules, addressing the regulatory documentation requirement that has been one of the primary adoption barriers for AI tools in registered investment advisory practices. "When Claude handles administrative tasks like meeting prep, follow-up drafts and pipeline review, advisers can spend more of their time serving clients," Anthropic said. "Practices can expand their customer base, and offer more of the services they need, like tax, estate, insurance and retirement planning." The launch arrives four days after OpenAI announced ChatGPT for Financial Services on September 10, initially focused on investment banking and equity research rather than the registered investment adviser segment that Anthropic is targeting with Claude for Financial Advisors. The two launches address adjacent but distinct parts of the financial services market, OpenAI targeting institutional investment banking workflows, Anthropic targeting the wealth management and adviser workflow layer, though both are competing for the same enterprise AI budget within financial services firms that operate across both segments. The human-in-the-loop architecture is an explicit design feature rather than a disclaimer. Adviser review and approval are required on critical tasks, meaning Claude for Financial Advisors is positioned as an automation tool that surfaces work for adviser sign-off rather than an autonomous agent that completes and sends client-facing outputs independently, according to the company. AI native startups Rogo and Hebbia operate in the same space, turning written instructions into financial models, memos, and pitch decks, and both run on Anthropic, OpenAI, and Google models. Anthropic's direct product launch into the adviser workflow layer places it in the same market as its own API customers.

## Adobe's AI-First ARR Is Up 150%. Firefly Grew 40% in One Quarter. The Stock Still Fell.

Date: 2026-09-15T10:37:48.127Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/adobes-ai-first-arr-is-up-150-firefly-grew-40-in-one-quarter-the-stock-still-fell

Adobe's AI metrics in Q3 FY2026 are the strongest the company has reported , according to the company. Revenue of $6.76 billion, up 13% year over year, beat estimates. Non-GAAP EPS of $6.13, up 15%, beat the consensus of $6.07. The company raised full-year guidance. And the stock still fell 2.14% after hours. The market reaction reflects a specific set of concerns that sit alongside the AI performance data. Remaining performance obligations grew 8% year over year, the first single-digit RPO growth since early fiscal 2023, which CFO Steve Day attributed to the company's deliberate freemium strategy and normal seasonal patterns rather than any change in underlying business momentum. Q4 RPO and CRPO typically step up in Adobe's seasonally strong enterprise quarter, and management said the pattern this year is consistent with prior years. The freemium strategy itself, prioritizing user acquisition before monetization conversion, is compressing near-term net new ARR even as it builds the long-term user funnel. AI-first ending ARR exceeded $650 million in Q3, up more than 150% year over year. Firefly ending ARR grew 40% quarter over quarter. Acrobat AI Assistant MAU doubled quarter over quarter. Creative freemium monthly active users crossed 100 million, up more than 70% year over year. Monthly active users across all Adobe businesses crossed 1 billion, growing more than 20% year over year. Each of Adobe Experience Manager, Adobe GenStudio, and Adobe Experience Platform delivered ending ARR growth of more than 20% year over year individually. The Agentic Software Thesis and the Chakravarthy Mandate The most strategically significant disclosure on the call was Anil Chakravarthy's framing of where Adobe is going. On his first earnings call as incoming CEO, Chakravarthy described agentic software as the next major transition for the software industry, directly analogous to the shift from on-premise to SaaS. He identified four specific things customers are looking for in agentic software: product architecture that supports any interface alongside existing Adobe functionality; model flexibility so customers can access the best model for each task; enterprise data and context so outputs are grounded and free of hallucination; and value realization support to make the full investment in agentic software commercially productive. "I am excited by the opportunity to drive Adobe's growth and leadership across creativity, productivity, and customer experiences in the era of agentic software," Chakravarthy said. The day before the earnings call, Adobe announced the Adobe Productivity Agent, a significant expansion

## OnTrac Appoints Arjun Sainath as CTO to Lead AI and Technology Strategy

Date: 2026-09-15T05:33:25.890Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/ontrac-appoints-arjun-sainath-as-cto-to-lead-ai-and-technology-strategy

OnTrac has appointed Arjun Sainath as Chief Technology Officer (CTO), putting a technology executive with more than two decades of experience in logistics, supply chain and enterprise technology in charge of its technology strategy. Sainath will oversee technology across operations, customer solutions, data, artificial intelligence and enterprise systems as OnTrac prepares for its next phase of growth. The carrier said his mandate includes building an AI-driven development model intended to accelerate speed-to-market and applying automation and advanced analytics across the business. The appointment comes as parcel carriers increasingly use AI to improve network efficiency, automate workflows and make operational decisions. UPS, for example, has expanded AI across logistics operations through digital twins, predictive planning and agentic control tower capabilities, reflecting the broader shift toward AI-enabled logistics infrastructure. Sainath Brings Logistics And Enterprise Technology Experience Before joining OnTrac, Sainath served as CTO at Cart.com, where he led product and engineering strategy and worked on AI and machine learning applications for commerce and fulfillment technology. Cart.com appointed him to the role in October 2025 as it expanded its investment in data-driven fulfillment systems. Sainath previously served as Corporate Vice President of Platform Engineering at Blue Yonder, where he led the migration of global supply chain technologies to a cloud-native software-as-a-service platform. Earlier in his career, he held technology leadership roles at Salesforce focused on mergers and acquisitions integrations and revenue automation, according to OnTrac. At OnTrac, his responsibilities will include the continued development of technology systems supporting the carrier's network. The company said it plans to use AI, automation and advanced analytics to improve network efficiency, support operational decision-making and develop new capabilities. The focus on data and operational intelligence mirrors a broader push among logistics companies to connect fragmented information and make it usable within day-to-day workflows. Penske Logistics, for example, recently launched an AI platform designed to unify transportation and warehousing data and provide real-time operational visibility. AI Becomes Part Of OnTrac's Growth Strategy Mike Brown, CEO of OnTrac, said Sainath's experience applying AI, data-driven solutions and modern platforms to operational challenges would help strengthen the company's technology foundation and accelerate innovation. Sainath said last-mile logistics offered an opportunity to use AI and automation to create customer value while lowering costs. His comments point to applications that extend beyond back-office automation into routing, network management and customer-facing delivery experiences. The carrier has already been developing technology around its network. On

## Copart Expands Beyond Salvage With AI and ACV Acquisition

Date: 2026-09-11T19:12:44.914Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/copart-expands-beyond-salvage-with-ai-and-acv-acquisition

Copart is expanding its vehicle remarketing business through software, data and a $1.9 billion acquisition of ACV, a digital automotive marketplace focused on dealer-to-dealer wholesale transactions. The deal, announced September 10, gives Copart access to ACV's dealer relationships, digital marketplace and vehicle-data technology while adding Copart's physical locations, logistics network and international buyer base. Copart expects the transaction to be breakeven when it closes and accretive to earnings in fiscal 2028, according to management. The acquisition comes as the economics of vehicle repair and insurance claims are changing. Rising repair costs, more complex vehicles and higher total-loss rates are affecting how insurers, dealers and auction operators assess and dispose of vehicles. AI Moves Closer to the Auction Floor For Copart, AI is being applied to more than back-office automation. Jay Adair, CEO and Executive Chairman of Copart, said the company was focused on using AI to connect buyers with vehicles. "The ability for a buyer to find that vehicle and to find a vehicle that matches their desired purchase is becoming more and more important every single day," Adair said during the company's fiscal 2026 earnings call. The company is applying technology across the buyer journey, from discovery and membership through bidding and purchasing. Adair also identified accuracy as a priority, saying automation can reduce errors in Copart's processes. Automotive companies are also implementing AI for vehicle inspection and vehicle-history data. AI-powered vehicle inspection is changing how cars are assessed, while Bumper's AI tool for vehicle history reports uses accident, maintenance, ownership and mileage information to help buyers evaluate vehicles. ACV brings similar capabilities into a wholesale marketplace. Its technology supports digital transactions, inspections, vehicle condition information and valuation. The company generated $760 million in revenue and $10.4 billion in marketplace gross merchandise volume in 2025, with 829,276 marketplace units transacted. Copart plans to keep ACV as a separate brand while integrating aspects of the two businesses. Adair said the company would evaluate ACV's technology for use across Copart and give buyers access to both businesses through connected experiences. Rising Repair Costs Change the Economics Copart's insurance business is operating against higher repair costs and changing claims patterns. Global insurance unit volume fell 4.2% in fiscal 2026, with U.S. insurance units down 7.5%. At the same time, Copart reported higher average selling prices and pointed to rising collision severity and total-loss frequency as industry trends. Average collision severity exceeded $6,300

## Cleveland Clinic Partners With Luminai to Transform Health System Operations

Date: 2026-09-11T11:09:54.514Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/cleveland-clinic-partners-with-luminai-to-transform-health-system-operations

Cleveland Clinic has partnered with Luminai, an AI-native enterprise automation platform for healthcare operations, to co-develop and deploy AI automation across complex administrative workflows. The collaboration will initially focus on external referrals, many of which still arrive by fax and require staff to sort, prioritize and transcribe information into the electronic medical record (EMR). Luminai’s system is designed to identify incoming referral documents, interpret their clinical context, match them with the appropriate patient and order type, and transcribe the information into the EMR for follow-up. Cleveland Clinic’s referral process continues to include fax submission for some physician referrals. The deployment builds on Cleveland Clinic’s use of Luminai for referral automation, extending the application of AI to a broader set of operational workflows. The earlier Cleveland Clinic-Luminai referral automation effort focused on reducing manual work involved in processing incoming referrals. AI Targets Complex Administrative Workflows Cleveland Clinic said the partnership will address processes that involve unstructured data, multiple hand-offs and complex decision-making. The health system plans to use the initial deployment to identify lessons that could inform automation of similar workflows elsewhere in the organization. “Automation only delivers value when it reflects the realities of clinical and operational work,” Rohit Chandra, Chief Digital Officer at Cleveland Clinic, said. He added that the collaboration was intended to reduce administrative burden while improving patient satisfaction. The referral workflow illustrates one of the challenges health systems face when administrative information arrives outside structured digital systems. Luminai said its platform combines healthcare-trained AI models with a configurable workflow engine and human-in-the-loop validation, allowing organizations to adapt workflows to their own operating requirements. Other health systems have also begun applying AI to administrative processes beyond clinical care, including referrals and operational workflows. Summa Health’s use of AI for referrals and other operations provides another example of health systems applying automation to work that traditionally requires staff to move information between systems. Luminai Engineers to Work With Cleveland Clinic Teams As part of the partnership, Luminai’s forward-deployed engineering team will work alongside Cleveland Clinic’s clinical and operational staff to adapt the automation workflows. The companies said this approach is intended to reduce the time between deployment and measurable operational impact. The broader objective is to apply the same approach to other administrative functions with similar characteristics. Cleveland Clinic and Luminai said the initial referral deployment could provide a basis for addressing workflows involving high volumes of unstructured

## DHL and Alibaba to Explore AI-Powered Logistics for SMEs Worldwide

Date: 2026-09-11T08:26:15.721Z
Author: Mukundan Sivaraj
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/dhl-and-alibaba-to-explore-ai-powered-logistics-for-smes-worldwide

DHL Group and Alibaba.com have signed a Memorandum of Understanding (MoU) to explore using artificial intelligence (AI) to simplify international logistics for small and medium-sized enterprises (SMEs). The companies will evaluate how AI-powered tools can support businesses across the international trade journey, including sourcing and supplier discovery, logistics quotations, bookings and shipment execution. The agreement was announced at Alibaba.com's CoCreate 2026 Conference in Los Angeles on September 10. The collaboration will focus initially on connecting DHL Global Forwarding's logistics services with Alibaba.com's Accio, an agentic AI platform designed to help SMEs execute business tasks without requiring code. "SMEs are the backbone of the global economy, but many still face challenges navigating the complex rules, regulations, and trade requirements that come with international expansion," Katja Busch, Chief Commercial Officer DHL and Head of DHL Customer Solutions & Innovation, said. "By bringing together Alibaba.com's digital commerce expertise and DHL's logistics capabilities, we want to explore how technology can help businesses spend less time managing complexity and more time focusing on growth." AI Agents Move Into Logistics Workflows Under the agreement, Accio will be connected to DHL Global Forwarding's Quotation & Booking capabilities. This would allow businesses to obtain real-time freight forwarding quotes, assess shipping options and book shipments through the AI platform. The approach reflects a wider shift toward embedding AI agents directly into operational workflows. For example, AI agents are being introduced into supply chain workflows, where they can interact with business processes rather than only provide information. DHL has also been developing AI applications across its logistics operations. Its existing digital tools include myDHLi, which provides customers with services such as freight quotations, bookings, shipment tracking, documentation and analytics. DHL's Quote + Book service allows customers to obtain air and ocean freight quotes and compare shipping options. The companies said the Accio integration would be the first in a planned suite of DHL logistics capabilities intended to automate business tasks and improve operational efficiency for SMEs. Kuo Zhang, President of Alibaba.com, said AI was making capabilities that previously required significant time, expertise and resources more accessible to SMEs. He said connecting Accio with DHL's logistics capabilities could help businesses move through cross-border commerce more quickly. Connecting Commerce With Freight Execution The partnership extends Alibaba.com's effort to make Accio useful across more stages of business operations. Alibaba launched Accio Work in March 2026 as a no-code enterprise AI agent for global

## CooperVision Turns to AI and Sales Expansion as It Rebuilds Growth

Date: 2026-09-11T06:21:00.778Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/coopervision-turns-to-ai-and-sales-expansion-as-it-rebuilds-growth

CooperCompanies is entering the final quarter of fiscal 2026 with the problem of converting demand into revenue. The medical device company reported $1.066 billion in fiscal third-quarter revenue, up 1% year over year and organically, while non-GAAP earnings per share rose 4% to $1.15. Free cash flow reached a record $273 million, and CooperSurgical delivered 3% organic growth. But CooperVision, the company's larger business, generated $717 million in revenue, essentially flat from a year earlier. The division's results were weighed by a deliberate reduction in U.S. channel inventory, while management said underlying U.S. consumption remained in the mid-single digits. CooperVision expects the inventory correction to continue into the fourth quarter. It expects Q4 revenue of $692 million to $706 million, representing organic growth of negative 2% to flat. The company is expanding its salesforce, increasing customer marketing and accelerating product development, while adding AI-driven targeting and analytics to its commercial operations. CooperVision's Growth Problem Is Moving From Inventory to Execution The inventory reduction explains much of the recent weakness, but Cooper CEO Al White also identifies commercial execution as an area where the company has struggled. "We don't have a manufacturing issue. Our team is strong. We're producing product. We don't have a distribution or logistics issue. We don't have a problem winning contracts," White says during the earnings call. "Where we've struggled is execution at the end of that, is actually executing and delivering revenues." CooperVision has continued to report strength in several parts of its portfolio. MyDay Toric and MyDay Multifocal delivered double-digit growth, while MiSight, the company's myopia-management lens, posted 20% organic growth in the quarter. The company also says it has won private-label contracts in the U.S. and Asia-Pacific that have not converted into revenue as quickly as expected. White says CooperVision has fewer salespeople in the field than some competitors and plans to expand its U.S. sales coverage by roughly 5,000 additional doors. New hires are expected to begin contributing more meaningfully during the second half of fiscal 2027. AI is being added to that commercial push. CooperVision says it is deploying AI-driven targeting and analytics tools alongside expanded sales coverage and customer marketing programs. The move reflects a shift in life sciences toward embedding AI into commercial operations, including targeting, field activity and customer engagement. The company expects the expanded salesforce and commercial investments to support growth in fiscal 2027. AI Becomes Part of

## Atlassian Says 80% of the Fortune 500 Now Use Rovo. Customers Using It Grow ARR Twice as Fast.

Date: 2026-09-11T04:05:46.833Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/atlassian-says-80-of-the-fortune-500-now-use-rovo-customers-using-it-grow-arr-twice-as-fast

Rovo, Atlassian's AI assistant and agent platform is no longer a pilot product at the Fortune 500. Atlassian CFO James Chuong confirmed at Citi's 2026 Global TMT Conference on September 9 that 80% of Fortune 500 companies now use Rovo. Rovo is enabled by default on Atlassian's paid Cloud plans, meaning the 80% figure tracks provisioning across the Fortune 500 install base rather than necessarily confirming daily active use at every organization. The more commercially precise signal is the ARR expansion rate. According to the company, customers using Rovo are growing their total Atlassian ARR at twice the rate of comparable customers not using it. That correlation of AI adoption doubling the expansion rate of the customers who have crossed the deployment threshold, is the clearest commercial validation Atlassian has provided for its AI strategy. It answers the question that most enterprise software AI disclosures leave unanswered: not whether customers are using the AI product, but whether using it is changing how much they spend on the platform. Rovo-assisted actions grew 50% quarter over quarter, a metric Chuong described as significant specifically because it measures task completion rather than simple search or chat. Atlassian disclosed at its Team '26 conference in May 2026 that Rovo supported more than 14 million assisted actions in a single month, giving the 50% quarter-over-quarter growth rate a concrete production baseline rather than a relative figure without context. "Users asking Rovo to take action," he said. "That's how you begin to unlock the power of AI across teams and across the board." The MCP Acceleration and the December Metering Decision Alongside the Rovo data, Chuong disclosed the scale of Atlassian's Model Context Protocol ecosystem. Monthly active users of Atlassian's MCP server have exceeded 1 million. MCP calls are growing 400% quarter over quarter. MCP-driven Jira item creation is up 4x and MCP-driven Confluence page creation is up 4x, both confirming that external AI agents are writing data back into Atlassian systems, not just reading from them. When an external coding agent creates a Jira ticket or a Confluence page as part of completing a task, Atlassian's system becomes the record of what that agent did, and the Teamwork Graph accumulates more context with every write, according to the company. The Teamwork Graph currently spans 150 billion connections across people, work, and data giving it the scale to function as a meaningful context engine for AI

## Battelle Study Finds AI Skills Alone Are Not Enough for Workforce Readiness

Date: 2026-09-10T14:44:36.056Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/battelle-study-finds-ai-skills-alone-are-not-enough-for-workforce-readiness

Artificial intelligence is changing what employers expect from new workers, but a new Battelle study found that AI proficiency alone does not define workforce readiness. The Applied Workforce Outlook, released September 9, surveyed 3,629 people across eight audience groups, including employers, human resources professionals, educators, students and parents. The study found a substantial gap between how employers and educators view preparation for work in an AI-enabled economy. Fewer than half of corporate and human resources decision-makers believed current curricula prepared students for real-world work. By comparison, 80% of high school educators and 85% of higher education educators believed their curricula prepared students for real-world application. The findings come as organizations and policymakers are placing greater emphasis on preparing workers for an economy where AI is changing job requirements. In June, RAISE US launched with a focus on aligning workforce training and education with an economy reshaped by AI. Employers Put Greater Weight on Human Skills Battelle’s research found that employers increasingly valued critical thinking, communication and adaptability alongside technical capabilities. Among corporate decision-makers surveyed, 96% rated communication skills as important, while 95% rated critical thinking and problem-solving as important. Adaptability was considered essential by 96% of corporate decision-makers and 98% of human resources leaders. The emphasis reflects a broader shift in how employers assess skills as AI becomes more capable of handling routine and information-heavy tasks. Battelle found that only 29% of corporate leaders believed AI-tool use alone defined job readiness. The World Economic Forum’s Future of Jobs Report 2025 similarly identified analytical thinking as the most widely valued core skill among employers, with 69% considering it essential. AI and big data ranked among the fastest-growing skills, alongside technological literacy, while resilience, flexibility, creative thinking and collaboration remained important. Companies are also experimenting with different approaches to developing these skills. Walmart, for example, announced plans to provide AI training across its U.S. and Canadian workforce, highlighting the role of worker preparation in AI adoption. Education and Employers Face a Readiness Gap The Battelle study also pointed to a disconnect between classroom preparation and practical experience. Thirty-three percent of high school educators identified an overly prescribed curriculum as a barrier to applied learning. The report argued that academic knowledge, work-based learning, project experience and AI fluency need to work together to prepare students for employment. It also identified communication, critical thinking, collaboration, adaptability, problem-solving and judgment as part of the broader

## Visa, Mastercard and Ant International Just Launched a Trust Layer for AI Agent Payments

Date: 2026-09-10T11:09:01.195Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/visa-mastercard-and-ant-international-just-launched-a-trust-layer-for-ai-agent-payments

AI agents that complete purchases on your behalf are useful precisely because they act without waiting for your confirmation at every step. That same autonomy is what makes them a trust problem for payment networks. Ant International, Mastercard, and Visa announced a framework on September 9, 2026 to address that problem at the infrastructure level rather than leaving it to individual platforms to solve independently. The Know-Your-Agent framework is designed so that an AI agent verified once across the network does not need to be verified again at every point of contact. "If an agent registers with Ant, they don't need to register again with Visa, Mastercard," said Jiang-Ming Yang, Chief Innovation Officer of Ant International. That single registration, covering identity, behavioral requirements, and ongoing monitoring, is the commercial efficiency the framework creates for agent platforms, marketplaces, and payment processors simultaneously. Visa, Mastercard, and Ant International have each already built their own agent trust protocols. The gap is not individual capability, it is interoperability across the systems that different consumers and merchants use. A consumer whose AI agent operates through Alipay+ needs that agent to be trusted by Visa-accepting merchants and Mastercard-processing banks without requiring a separate verification journey at each point. According to the press release, the KYA framework rests on three specific requirements. Cross-network operator traceability links each agent to a validated operator, cardholder, or business, establishing clear attribution for every agent action. Shared certification requirements assess each agent against security and behavioral standards before it is permitted to transact. Continuous transaction monitoring evaluates each agent using identity and transaction signals on an ongoing basis rather than treating certification as a one-time gate. The framework will be developed through BuildFin.ai, an industry platform convened by the Monetary Authority of Singapore, giving it regulatory grounding in one of the world's most developed AI financial governance frameworks. The MAS backing also positions the initiative as an Asia-Pacific infrastructure effort rather than a bilateral commercial arrangement between three companies. Digital wallets accounted for 56% of global e-commerce value in 2025, more than $13 trillion in spending per Worldpay, and Ant International's Alipay+ system connects more than 50 e-wallet partners across markets where digital wallets are the primary payment method rather than physical cards. A KYA framework that spans both card networks and wallet ecosystems addresses the full commercial surface on which agentic commerce will operate. "Interoperability across Know-Your-Agent frameworks is essential

## UF Health Partners with R1 to Deploy AI-Powered Revenue Cycle Platform

Date: 2026-09-10T11:01:10.127Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/uf-health-partners-with-r1-to-deploy-ai-powered-revenue-cycle-platform

UF Health has partnered with healthcare revenue management company R1 to deploy an AI-native operating model for revenue cycle management (RCM), the companies announced September 9. The collaboration will use R1’s Phare OS platform and its R37 Innovation Lab to support revenue cycle operations across the University of Florida Health system. UF Health’s network includes 11 hospitals and hundreds of physician practices and outpatient locations across Florida. The initiative comes as healthcare providers increasingly apply AI to administrative and financial workflows. AI has already moved into areas such as claims processing, denial management and revenue recovery, with providers looking to automate work that remains labor-intensive. Our recent look at AI in healthcare revenue cycle operations found that specialized AI systems are being used for tasks such as eligibility verification and denial management, with human review retained for higher-risk cases. R1 Brings Phare OS Into Academic Health System At the center of the UF Health collaboration is Phare OS, which R1 launched in 2025 as a platform intended to connect workflows across the revenue cycle rather than address individual processes in isolation. R1 said Phare OS is built on three core components: its Data Platform, Payer Atlas and Phare Intelligence. The company said Payer Atlas has more than 1,500 payer connections and processes more than 600 million payer transactions annually. The platform was live across provider organizations representing more than $76 billion in net patient revenue as of 06/18, according to R1. The company has continued expanding Phare OS through capabilities covering areas including denials management and prior authorization. R1 said engineers from R37 will work directly with UF Health teams. The model combines agent-driven workflows with human governance, allowing staff to oversee AI-supported processes rather than removing human involvement from the revenue cycle. That approach reflects a broader shift toward AI systems that execute defined administrative workflows while retaining human oversight. Healthcare technology company TELCOR, for example, has also incorporated AI for prior authorizations, appeals, payer follow-up and document processing with human oversight. Focus on Operational and Financial Performance UF Health CFO Geoff Gardner said the health system selected R1 because of its combination of revenue cycle expertise and AI capabilities. “We chose to collaborate with R1 because of its ability to pair 20 years of deep revenue cycle expertise with advanced AI capabilities for revenue management,” Gardner said in the announcement. The collaboration is intended to identify opportunities to

## Rentosertib Shows Potential to Lower Biological Age in Small AI Drug Study

Date: 2026-09-10T08:56:14.288Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/rentosertib-shows-potential-to-lower-biological-age-in-small-ai-drug-study

A drug developed with artificial intelligence (AI) to treat idiopathic pulmonary fibrosis (IPF) appeared to lower biological age markers in a small group of patients, according to a study published September 7 in Nature Biotechnology. The researchers analyzed blood samples from 42 people with IPF who participated in a previous Phase 2a clinical trial of rentosertib, an experimental drug developed by Insilico Medicine. They applied six proteomic aging clocks to the samples and found that all six predicted lower biological age among patients who received the drug over the 12-week study period. The finding adds a new dimension to rentosertib's development. The drug was originally developed to target IPF, a progressive lung disease in which scar tissue builds up in the lungs. Its discovery involved generative AI identifying Traf2- and Nck-interacting kinase (TNIK) as a potential disease target and helping design a molecule against it. The development is part of a broader push to move AI-discovered drug candidates from computational research into clinical testing, a transition explored in recent work on AI drug discovery reaching clinical reality . Six Aging Clocks Pointed to Lower Biological Age The researchers used proteomic measurements covering 2,841 proteins from serum samples collected during the Phase 2a trial. The six aging clocks used different models and training approaches, including systems designed to estimate chronological age and others trained to assess mortality risk. Across the treatment groups, the clocks consistently recorded reductions in predicted biological age, while the placebo group showed minimal change or slight increases. The strongest overall signal appeared in the 30 mg twice-daily group, with nine statistically significant reductions across the clock comparisons. The study also identified changes in proteins and biological pathways associated with cellular senescence and metabolism, alongside changes linked to fibrosis. However, the researchers cautioned that the proteomic clocks could not distinguish whether the changes reflected an effect on aging itself or improvements related to the underlying lung disease. That distinction is important because biological age in the study was a calculated measure derived from protein patterns, rather than evidence that patients' chronological age had been reversed or that they would live longer. The findings also build on Insilico's broader work applying AI to aging research, including efforts to develop AI models that can analyze biological and clinical data for longevity applications. That work has increasingly connected AI drug discovery with aging research. Clinical Testing Still Focused on Lung Disease

## SailPoint’s AI-Driven ARR Surpasses $70 Million as Agentic Security Demand Grows

Date: 2026-09-10T05:39:32.210Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/sailpoints-ai-driven-arr-surpasses-70-million-as-agentic-security-demand-grows

SailPoint’s AI business crossed $70 million in annual recurring revenue (ARR) in the second quarter of fiscal 2027, as demand for tools that govern artificial intelligence agents helped drive more than 30% of the company’s net new ARR. The identity security company reported total ARR of $1.231 billion for the quarter ended 07/31, up 25% year over year. SaaS ARR rose 36% to $847 million, while total revenue increased 17% to $309 million. SailPoint’s AI-driven pipeline also more than doubled to over $200 million since its June Investor Day. The company said customers that adopted an AI-driven solution increased their annual spending by more than 60%, while more than two-thirds of SaaS migrations during the quarter included an AI-driven solution. The results come as enterprises expand the use of AI agents that can access data and systems without direct human intervention. The shift has created a new identity-security challenge alongside the broader expansion of enterprise AI. For example, Okta’s AI-agent push has also focused on managing identities and access as autonomous systems become more common. SailPoint Expands Agentic Identity Security SailPoint has been building its AI strategy around the idea that enterprises need to govern both human and non-human identities through a common security framework. In August, the company launched its SailPoint Identity Security solution, combining Agentic Fabric for AI agents and other non-human identities with Human Fabric, the evolution of its Identity Security Cloud. Agentic Fabric was made generally available as part of the launch. During the earnings call, CEO Mark McClain said SailPoint’s AI-driven solutions accounted for more than 30% of net new ARR in Q2. He also said the company had seen growing demand from customers seeking to understand which agents exist in their environments, what those agents can access and which humans are responsible for them. That focus puts SailPoint into a cybersecurity market increasingly concerned with continuous identity controls for autonomous systems, a theme also emerging in CrowdStrike’s approach to continuous identity security for AI agents . SailPoint also completed its acquisition of Entro Security during the quarter. Entro added technology for discovering and securing non-human identities and credentials, including more than 1,200 types of secrets, tokens and certificates across enterprise environments. The company said the acquisition had contributed less than $3 million in ARR during the quarter, with currency headwinds more than offsetting that contribution. SaaS Transition Creates Revenue Timing Impact SailPoint’s stronger ARR

## Figma Cut Its AI Tool Pricing by 50% to Drive Volume

Date: 2026-09-09T10:44:28.456Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/figma-cut-its-ai-tool-pricing-by-50-to-drive-volume

Figma cut the price of Figma Make credits by approximately 50% in some cases, and CEO Dylan Field was direct about why at the Goldman Sachs Communacopia + Technology Conference on September 8. "We did the gut check, invariant of customer feedback, and went, this feels a bit too high, and it's most important to get the usage of the system and the volume growing," Field said. Usage rose after the reduction. Field declined to disclose specific figures but confirmed the direction. Figma is prioritizing volume over near-term margin, betting that falling frontier model token costs will improve the economics over time while usage scales, according to the company. The same trade-off underlies Figma's broader AI product strategy. Field described Code Layers, announced at Config and not yet launched, as the same concept as Figma Make in a different environment. Where Figma Make creates applications from a prompt, Code Layers places working code directly on the canvas next to design elements, allowing teams to prototype, design, and view code simultaneously in one environment. Field said Code Layers generated strong customer demand before launch. "Everyone saw that and went, 'I want it now,'" he said. Field's central argument at the conference was that the canvas is the durable interface for product creation work. External coding agents are increasingly writing changes back to Figma rather than operating independently, which Field described as commercial validation of that thesis. The View Figma tool call, which enables that write-back, is growing quarter over quarter, according to the company. Figma Agent is being designed to extend the same multiplayer technology that allows multiple humans to collaborate on a single canvas to AI agents, bringing agents and designers into the same environment simultaneously. Field named speed, cost, and quality as the three goals for the agent and acknowledged the company is still working toward all three. On competitive positioning, Field said frontier model progress on design remains incremental. Visual quality, UX details, design-system adherence, and hallucination are all unresolved, and the hallucination problem in particular requires positive and negative training examples from customers rather than written guidelines alone. "Strict adherence to these guidelines is not a given," Field said. "It's an easy thing to say and a harder thing to actually make happen." Figma's gross margin for the next six months is uncertain. Field said usage and monetization patterns for Code Layers and Figma Make are still

## Accenture and Google Cloud Launched a Dedicated Gemini Enterprise Business Group

Date: 2026-09-09T10:42:15.473Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/accenture-and-google-cloud-launched-a-dedicated-gemini-enterprise-business-group

Accenture and Google Cloud launched a dedicated business group for Gemini Enterprise deployments on September 8, and the most commercially specific detail in the announcement was of YouTube. During NFL Sunday Ticket surge demand, YouTube deployed a Gemini Enterprise agent with Accenture and Google Cloud. Customer sentiment improved 11%. Average handle time fell 37%, according to the press release. The result: customer sentiment up 11%, average handle time down 37%. The business group itself is structured around scale. It draws on Accenture's existing base of nearly 50,000 Google Cloud-skilled professionals and adds a 1,000-person Forward Deployed Engineer workforce, engineers embedded directly with clients during implementation rather than operating from a distance. The FDE model is designed to compress the gap between AI deployment and production outcomes that has driven the broader enterprise AI ROI problem documented throughout 2026, according to the press release. "Deploying agentic AI is a top priority for enterprises today, and the Accenture Gemini Enterprise Business Group significantly expands the expertise and resources available to help our customers deliver real business value," said Thomas Kurian, CEO of Google Cloud. The business group operates across four priority areas, according to the press release. The first is adoption, deploying Gemini Enterprise faster through proprietary accelerators and implementation frameworks purpose-built for the platform. The second is industry solutions, building repeatable, sector-specific AI deployments that reduce the time from implementation to measurable value. The third is capability centers, dedicated infrastructure designed to help enterprises move beyond AI experimentation into production-scale transformation. The fourth is user adoption, ensuring that Gemini Enterprise capabilities are used at scale once deployed rather than remaining confined to pilot teams. The group builds on Accenture and Google Cloud's existing Generative and Agentic AI Center of Excellence and the Gemini Enterprise Acceleration Program, two joint structures that predate the new business group and represent the implementation foundation on which it is built. Accenture has been named Google Cloud's Global Services Partner of the Year for four consecutive years, giving the partnership a documented track record that the new group is designed to accelerate rather than establish from scratch . "The Accenture Gemini Enterprise Business Group will help clients achieve these outcomes faster, bringing together the talent, advanced AI and data capabilities, and industry expertise needed to reinvent with confidence and create value at scale," said Julie Sweet, Chair and CEO of Accenture.

## Truist Has Delivered 2 Billion AI-Powered Financial Insights Since 2023

Date: 2026-09-09T09:33:22.196Z
Author: Sachin Mohan
Desk: Banking
URL: https://aimmediahouse.com/banking/truist-has-delivered-2-billion-ai-powered-financial-insights-since-2023

Two billion proactive financial insights delivered to clients since 2023. That is the number Truist Financial Corporation announced on September 3. Truist Insights uses data, analytics, and AI-powered personalization to turn a client's financial activity into timely, contextual observations. The system monitors cash flow patterns, income and expense changes, subscription activity, potential duplicate transactions, and balance shifts, and surfaces those observations inside the digital channels where clients are already engaging, including account details, planning dashboards, Truist Assist, and push notifications. The integration is designed so that a client managing their account sees a relevant observation in context rather than receiving a separate alert requiring them to navigate to a different part of the app, according to the company. "Reaching 2 billion insights represents 2 billion moments of proactive care, opportunities to deliver knowledge and support in ways that are personal, timely and useful," said Dontá Wilson, Chief Consumer and Small Business Banking Officer at Truist. "By combining AI-driven, proactive insights with the expertise of our teammates, we're creating experiences that are digitally empowered and deeply relational." Truist Insights carries a 4.5 client satisfaction rating, and the company measures engagement not by opens alone but by whether a client takes the next step after viewing an insight, such as navigating to a recommended product or viewing additional detail. That behavioral signal is more commercially meaningful because it measures whether the insight prompted action rather than merely capturing attention . More than 70% of overall Truist transactions now happen digitally and more than 80% of deposits occur through self-serve channels, according to the company. At that level of digital engagement, an AI personalization layer that surfaces relevant observations in the flow of the digital banking experience reaches a meaningful share of the client base without requiring them to seek out the feature separately. "Personalization is how we show care at scale," said Sherry Graziano, Head of Digital, Client Experience and Marketing at Truist. "By using data responsibly to anticipate client needs and connect them to the right support at the right time, we're making banking more personal, proactive and relevant in the moments that matter." Truist said it is continuing to expand Truist Insights into more channels and teammate tools, using client feedback, research, and design practices to improve digital journeys while maintaining the connection between AI-driven personalization and human support. Clients retain the ability to opt out of Truist Insights at

## Enterprises Are Scaling AI Before They Are Ready For It

Date: 2026-09-09T07:55:48.542Z
Author: Deployed
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/enterprises-are-scaling-ai-before-they-are-ready-for-it

Companies are no longer debating whether they should use AI. Boards are asking how quickly it can be deployed, how many processes can be automated and how much money it can save. Employees are experimenting with copilots, agents and vibe coding, while technology teams are under pressure to turn these experiments into production systems. Yet, according to data and AI leaders across industries, enterprises may be making the same mistake at scale: they are trying to accelerate AI before fixing the foundations that make AI useful. The gap is no longer between companies that have AI and those that do not. It is between companies that have demonstrated what AI can do and those that are actually ready to run it reliably across the business. Dr. Santosh Karthikeyan Viswanathan , Data and AI Leader, says the problem starts at the top. “Leaders are under pressure in boardroom discussions to scale AI and it’s on their scorecard. But without building AI literacy for leaders first, enterprises struggle to channel budget to the right projects with measurable ROI and real business value.” That creates a familiar enterprise paradox. The board wants an AI strategy, the CIO wants an AI roadmap and business teams want AI-powered products. But if leadership does not understand where AI can realistically create value, investment can quickly become a collection of disconnected pilots. The result is an organisation that looks busy with AI without necessarily becoming better at using it. The next problem sits deeper in the stack: data. For industrial enterprises in particular, simply having large amounts of data does not mean an organisation is AI-ready. Guru Ananthanarayanan , Managing Director, India at Cognite, argues that the real challenge is connecting information that was never designed to work together. “From an Industrial AI perspective, AI isn't the hard part. The real challenge is bringing together data from IT systems, operational technology (OT) like machines and control systems, and engineering information (ET). On their own, these data sources don't naturally connect.” This is particularly important in industries where AI has to understand the physical world. A factory may have sensor data from a machine, maintenance records sitting in another system and engineering drawings stored somewhere else. An AI model may technically be capable of analysing all three, but unless the enterprise can establish the relationship between them, the output may have little meaning to the person operating the

## Adobe's CEO Search Took Six Months and Ended With an Internal Hire

Date: 2026-09-09T04:40:17.920Z
Author: Sachin Mohan
Desk: Retail
URL: https://aimmediahouse.com/retail/adobes-ceo-search-took-six-months-and-ended-with-an-internal-hire

Adobe announced its next CEO on September 3, 2026. Anil Chakravarthy will become President and CEO effective December 1, succeeding Shantanu Narayen who announced in March he would step down after 18 years at the helm. The Chakravarthy appointment is an internal hire. He currently leads Adobe's Customer Experience Orchestration and Worldwide Field Operations and joined the company approximately seven years ago from Informatica, where he served as CEO. Narayen will move to Executive Chair and work closely with Chakravarthy through the transition. On the same day, David Wadhwani announced he would leave Adobe. Wadhwani has led Adobe's Creativity and Productivity business, the division that houses Photoshop, Premiere, and Illustrator, for nearly five years. CNBC had previously reported he was seen as a strong contender for the CEO role, in part because of his central role in Adobe's $20 billion bid to acquire Figma, which was ultimately abandoned in December 2023 due to regulatory hurdles. Wadhwani said in a LinkedIn post he has company ideas he plans to explore and described the moment as "the right time to start a new chapter." "I've had the privilege to work alongside Anil the last few years and I leave confident in where he'll take Adobe," Wadhwani wrote. The Search That Ended With an Internal Hire When Adobe announced in March that Narayen would step down, the company said it would consider both internal and external candidates. The six-month search concluded with an internal appointment. "If this was Plan A, what took so long?" KeyBanc Capital Markets analyst Jackson Ader wrote. "The awkward implication here is that this wasn't Plan A or maybe even Plan B and the Company had a hard time attracting the kind of outside talent befitting a company of Adobe's stature." William Blair analyst Arjun Bhatia offered a different reading of the same outcome. "The decision to go internal shows Adobe values continuity in leadership and is confident in its current strategy and vision," Bhatia wrote. Adobe shares fell 6.1% on September 4 to $268.36. The stock is down 23% year-to-date, compared with a 13% gain for the S&P 500. Adobe specifically faces the risk of AI-native creative tools reducing the switching costs that have historically made its Creative Cloud subscription base durable. Chakravarthy's Profile and the AI Mandate Chakravarthy's background is in enterprise data management and customer experience. He joined Adobe in 2019 after serving as CEO of

## Qualderm Partners With Parakeet Health to Expand AI-Powered Patient Access

Date: 2026-09-08T16:24:25.582Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/qualderm-partners-with-parakeet-health-to-expand-ai-powered-patient-access

Qualderm, a national physician-led dermatology network supporting nearly 160 practices across 17 states, has partnered with healthcare AI company Parakeet Health to use agentic AI for appointment scheduling, cancellation recovery and patient follow-up. Under the partnership, Parakeet’s platform will evaluate provider schedules and practice-level requirements to identify available appointment capacity. It will then reconnect patients who canceled visits, missed appointments or are overdue for follow-up care. The companies announced the partnership on September 8 as dermatology practices face growing demand for care and pressure on clinical and administrative staffing. The deployment is designed to help Qualderm make greater use of appointment capacity while reconnecting patients with care. AI Targets Unused Dermatology Capacity Qualderm’s network spans medical and surgical dermatology, skin cancer care, cosmetics and plastic surgery. Its size means scheduling can involve different providers, services and practice-level requirements. Parakeet’s platform handles inbound and outbound patient communications across voice, SMS, fax and email. Its capabilities include appointment scheduling, referral conversion, cancellation recovery and patient recall. The company said its system can account for provider- and practice-level scheduling rules when determining appropriate appointment options. That allows the AI to work across different workflows rather than treating every available calendar slot the same way. The approach reflects a broader push to apply AI to healthcare access and administrative workflows. Assort Health’s agentic AI platform , for example, has focused on automating patient interactions including scheduling and referrals. Jung Park, CEO and Co-founder of Parakeet Health, said specialty care requires AI systems to account for different services, providers, scheduling rules and patient needs. Parakeet also said its platform is designed to operate within existing electronic health record and practice management workflows while supporting deployments across large networks. Healthcare AI Moves Into Patient Access The Qualderm partnership comes as healthcare organizations increasingly apply AI to operational functions beyond clinical care. Summa Health’s use of AI across referrals, patient access and other operations reflects the expansion of specialized AI into healthcare workflows. Parakeet reported 38% more filled appointments in head-to-head pilots with two legacy vendors, along with a 6x increase in outbound conversion. The company also reported 42% more calls answered and a 60% reduction in call-center operating expenses. These figures are company-reported results. The company uses a pay-for-performance model, charging based on outcomes such as answering patient questions, rescheduling appointments and canceling appointments. Parakeet said its pricing is tied to results rather than software seats.

## Adventist Health Launches Epic System With Expanded AI Integration

Date: 2026-09-08T07:25:49.502Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/adventist-health-launches-epic-system-with-expanded-ai-integration

Adventist Health completed the largest technology implementation in its history, connecting 27 hospitals and more than 440 clinics across California, Oregon and Hawaii through a single Epic electronic health record (EHR) system. The health system said the deployment is designed to give clinicians a unified view of patient information across more than 100 communities. Adventist Health manages more than 5 million patient encounters and 3 million clinic visits each year. The implementation also includes an expanded deployment of Epic’s artificial intelligence capabilities. Adventist Health said it was the first U.S. health system to go live with this level of Epic AI integration in a single implementation, using more of the platform’s AI-powered tools at launch than any U.S. health system previously had. The claim was made by Adventist Health and was not independently quantified in the announcement. Epic AI Moves Into Clinical Workflows The AI deployment is focused in part on reducing documentation and administrative work for clinicians. Adventist Health said AI would support caregivers while leaving decisions about patient care with human clinicians. That approach follows a broader move by health systems to embed AI within existing EHR workflows. Providence has expanded its use of Epic-native AI tools , including applications for clinical documentation summaries, coding assistance and reporting. Sutter Health has also integrated AI directly into Epic , allowing physicians to access clinical evidence through natural-language queries within the EHR. Epic has continued expanding its own AI capabilities across clinical workflows. Its recent deployments include Chart with Art for nursing documentation at Mount Sinai Medical Center and AI-supported Ergo Visit at Ochsner Health. Adventist Health President and CEO Kerry L. Heinrich said the unified record would allow patient information to follow people across different care settings while AI handles some of the technology-enabled workload. “When a patient’s full story lives in one place, we can care for them more safely and with deeper compassion,” Heinrich said. The health system also said patients would be able to connect with their care teams through a single MyChart account. Unified Record Connects Care Across 100 Communities The Epic implementation is intended to give care teams access to the same patient record as people move between clinics, hospitals and home-based care. Adventist Health said this could support continuity and coordination by making relevant information available across its network. The patient-facing system also brings appointments, communications and other health information into a single

## OSF Expands AI Stroke Assessment to Rural Hospitals Across Its Network

Date: 2026-09-08T06:37:30.834Z
Author: Mukundan Sivaraj
Desk: Pharma
URL: https://aimmediahouse.com/pharma/osf-expands-ai-stroke-assessment-to-rural-hospitals-across-its-network

OSF HealthCare is expanding the use of artificial intelligence (AI) for stroke assessment across its hospital network, including rural facilities and OSF St. Joseph Medical Center in Bloomington, Illinois. The health system has expanded RapidAI from select hospitals to all 18 hospitals in its Illinois and Michigan network. The platform analyzes medical images and provides automated measurements and visualizations that can help clinicians assess suspected strokes and coordinate treatment. The expansion follows OSF's earlier deployment of RapidAI at OSF HealthCare Saint Francis Medical Center in Peoria and OSF HealthCare Saint Anthony Medical Center in Rockford. The technology is now being extended to hospitals where access to neurologists and advanced stroke services can be more limited. OSF's expansion is part of a broader move toward AI-powered tools becoming part of routine hospital workflows , particularly for time-sensitive imaging tasks. AI Adds Blood-Flow Analysis to Stroke Imaging RapidAI can analyze CT scans and CT perfusion images to provide information about blood flow in the brain. OSF said the system gives clinicians quantitative measurements that can help distinguish the area already affected by a stroke from tissue that may still be at risk. Arun Talkad, MD, Director of Stroke Care at OSF HealthCare, said the technology can reduce the need for physicians to estimate the extent of affected brain tissue from images alone. The system also allows neurologists to review stroke imaging through mobile devices, according to Talkad. That can allow specialists to begin reviewing scans while a patient is still moving through the emergency department rather than waiting for images to be reviewed at a specialist's workstation. The emphasis is significant because treatment decisions for stroke can depend on how much brain tissue has been damaged and how much remains viable. OSF said its initial RapidAI deployment produced improvements in treatment times and more consistent identification of patients who could require stroke intervention. The technology is also intended to help identify hemorrhagic strokes, which involve bleeding in the brain. In rural hospitals without neurosurgical or intensive-care capabilities, faster identification can help clinicians determine whether a patient needs to be transferred to a larger medical center. Expanding AI Across OSF's Stroke Network OSF's systemwide deployment builds on its earlier use of AI-assisted perfusion imaging. In 2024, OSF HealthCare Little Company of Mary Medical Center began offering CT and MRI perfusion capabilities using RapidAI to analyze blood flow and provide clinical information to physicians.

## Google DeepMind's WeatherNext 3 Is Now Inside Search, Maps and Gemini

Date: 2026-09-08T05:10:37.972Z
Author: Sachin Mohan
Desk: Manufacturing
URL: https://aimmediahouse.com/manufacturing/google-deepminds-weathernext-3-is-now-inside-search-maps-and-gemini

WeatherNext 3 is live across Google's consumer products as of September 3. This is the first time that Google DeepMind's AI weather forecasting model has directly powered the weather information users see in Google Search, Google Maps, and the Gemini app simultaneously. The model has a measurable advantage over WeatherNext 2. Resolution improved from a 25-kilometer grid updated every six hours to 5-kilometer resolution updated every hour, a five times improvement in spatial sharpness combined with a six times improvement in update frequency, according to the company. The leap is enabled by a fundamental change in what the model learns from: WeatherNext 3 ingests live geostationary satellite data on an hourly basis, bypassing the 6-hour data lag that traditional numerical weather prediction systems carry. Most AI weather models, including WeatherNext 2, train on data produced by those traditional systems, inheriting their lag as a result. WeatherNext 3 trains directly on raw satellite observations, giving it a more current view of the atmosphere at the moment a forecast is generated. "Weather is chaotic, and so small differences really start to perturb massively," said Ferran Alet, Staff Research Scientist Manager at Google DeepMind. "Machine learning targets the problem we are really solving, which is approximate noisy physics from incomplete information and finite compute, and so it learns patterns from a lot of data." The model also forecasts 100-meter wind speeds, roughly at turbine height, for wind energy output planning, alongside high-resolution cloud cover and solar radiation estimates for solar farm production projections, according to the company. Both are designed to give grid operators and renewables developers the ability to predict how much power their clean energy assets will generate and match it with consumer demand. The clean energy variables connect directly to enterprise deployments already disclosed this year. Ryanair's five-year Google Cloud partnership announced in August 2026 specifically named WeatherNext as one of the Google DeepMind models being deployed for fleet operations and maintenance scheduling. On the performance benchmark, WeatherNext 3 placed first on Operational WeatherBench, the independent AI forecast comparison utility built by Brightband, ahead of deep-learning models from Microsoft and NVIDIA and ahead of traditional physics-based forecasts from both the US National Weather Service and the European Centre for Medium-Range Weather Forecasting. Precipitation forecasting improved up to 60% against NASA's IMERG satellite dataset. Forecast data is available to researchers and developers through BigQuery, Earth Engine, and Google Cloud Storage. WeatherNext