JPMorgan's AI Chief Steps Down Just as the Bank Deploys What She Built.

Teresa Heitsenrether joined JPMorgan Chase as a summer intern in 1987. Thirty-nine years later, she announced that she is stepping down at the end of 2026 as the executive who built one of the most advanced AI infrastructures in global banking. 

What's notable is that she’s departing at the precise moment the bank is preparing to deploy what she spent three years constructing.

The announcement came in a memo from CEO Jamie Dimon and COO Jennifer Piepszak on July 1, 2026. "Teresa has played a pivotal role in building and transforming some of our most important institutional businesses and, most recently, in shaping the firmwide data and artificial intelligence strategy that is central to our future," they wrote. 

Heitsenrether, now 61, currently serves as Chief Data and Analytics Officer and sits on JPMorgan's 12-person operating committee, the bank's highest-level decision-making body below the board.

Her successor is Scot Baldry, JPMorgan's Chief Technology Officer. Baldry will absorb Heitsenrether's title and responsibilities, but he will not join the operating committee. He will report to Chief Information Officer Lori Beer, who retains her own committee seat. 

The AI strategy that had its own direct committee-level voice will now be housed within the technology organization, one level removed from where it has sat for the past three years, according to the company.

What Three Years of AI Strategy Built

Heitsenrether's tenure leaves a record of named AI systems with documented deployments. LLM Suite is the most wide-reaching of those systems. More than 230,000 JPMorgan employees now use it to draft reports, automate compliance processes, identify fraudulent activity, analyze markets, and support customers. 

Employees using LLM Suite have reported efficiency gains of 30-40%, according to the company. The platform gives workers across the enterprise access to agentic tools, data, and systems wherever they are needed rather than only where IT has deployed them.

Contract Intelligence (COiN) automated the analysis of legal documents that would otherwise have taken 360,000 person-hours to complete manually. CoachAI provides real-time advice to JPMorgan's wealth managers and EVEE is the bank's intelligent call center assistant.

A proprietary coding tool is available to software engineers for tasks specific to JPMorgan's infrastructure, including migrating legacy systems to newer architecture.

Together, these systems have saved JPMorgan an estimated $2 billion annually as of 2025, according to Dimon's own disclosure. The revenue impact is also becoming measurable in client-facing roles. 

JPMorgan has disclosed a 20% increase in private banking gross sales attributed to AI, and has said AI could eventually allow bankers to expand their client coverage by as much as 50%.

The bank has topped Evident AI's index of AI maturity at financial institutions for four consecutive years, not by deploying chatbots on existing services, but by treating AI as an organization-wide transformation built on connected infrastructure, according to the company.

The most consequential partnership Heitsenrether drove was JPMorgan's founding membership of Anthropic's Project Glasswing, the program giving a select group of trusted organizations access to Anthropic's Mythos model. 

JPMorgan is one of the few financial institutions to have secured that access, and Heitsenrether negotiated its position in that group during her tenure.

The $19.8 Billion Budget and What Comes Next

In February 2026, JPMorgan announced a technology budget of $19.8 billion for the year, a roughly 10% increase from 2025, making it one of the largest single-year technology investments by any financial institution in history. 

CFO Jeremy Barnum called technology "a major driver of expense growth" and noted that generative AI is growing as a proportion of the bank's total AI usage.

Dimon acknowledged at the same presentation that returns on AI are "uniquely hard to measure" — initiative by initiative, the ROI is difficult to quantify, and time saved is often "too vague" to express concretely. His answer to that ambiguity was not to slow the investment. 

"We need to have the best tech in the world," he said. "That drives investment, it drives margin, it drives competition."

The most consequential step in that investment is still ahead. Derek Waldron, JPMorgan's Chief Analytics Officer, told CNBC in a June 9, 2026 exclusive that the bank plans to deploy a new generation of AI agents later this year that can operate autonomously for far longer than current systems. 

"We've entered now the era of long-running autonomous agents," Waldron said. "That means that agents don't just run for two or three minutes to carry out a goal or some instructions of a human, they can run for an hour or two." 

He confirmed deployment in 2026 and described the agents as capable of coordinating complex workflows across multiple software environments, not single-step task completion. 

Waldron also said he expects the technology to continue advancing beyond hour-long sessions: "Eventually, AI agents will remain coherent for multiple hours, then days, then weeks."

These are not tools that respond to employee requests. They are systems that will operate independently within defined workflows, making decisions and taking actions without a human initiating each step, and JPMorgan's planned deployment, Waldron told CNBC.

Heitsenrether built the infrastructure that makes that deployment possible. Baldry will execute it.

The Headcount Question

Behind the infrastructure announcements and the technology budget sits a workforce question that JPMorgan has been unusually direct about. 

In 2025, the bank told investors that AI would enable at least a 10% headcount reduction in operations and account services, and described that estimate as conservative. Dimon has said that AI will eliminate jobs, and that anyone who disagrees should "stop sticking their heads in the sand."

The reality so far has been more nuanced. Headcount has remained broadly steady, with reductions in administrative and support functions offset by growth in client-facing and technical roles. 

The bank is also considering a structural change to its junior banker pipeline, reducing the ratio of junior to senior bankers from 6:1 to 4:1, which would allow senior bankers to spend more time mentoring while AI handles a greater portion of the analytical work juniors currently perform.

The bank frames workforce displacement as a strategic objective to be managed rather than a side effect to be minimized. 

Retraining and redeployment programs are specifically designed to retain the experience and institutional knowledge the bank has spent decades developing, on the grounds that losing experienced people and retraining their replacements from scratch is more expensive than managing the transition proactively.

Heitsenrether's departure comes in the context of broader leadership change at JPMorgan. The bank named Doug Petno and Troy Rohrbaugh co-presidents last week, positioning both as leading candidates to eventually succeed Dimon, while announcing the retirement of Marianne Lake, long seen as a front-runner for the top job. 

Dimon, who turned 70 this year, has told the board's succession planning process that multiple executives are capable of eventually becoming CEO.

What Heitsenrether leaves behind is an organization that has spent three years building the governance, data, and infrastructure layer that autonomous agents require to operate safely inside a regulated financial institution. 

Waldron's description of where this goes next  is the clearest public articulation of what JPMorgan is building toward.