AIG and McGill & Partners Use Agentic AI to Deploy Insurance Capacity in Real Time
"This collaboration has the potential to disrupt the dynamics of the subscription market"
On March 16, 2026, AIG and McGill & Partners announced a long-term strategic collaboration that will use AI to manage the deployment of insurance capacity across McGill's specialty portfolio in real time.
Under the agreement, AIG will provide 25% capacity across up to $1.6 billion of McGill & Partners' gross premiums written specialty portfolio, with underwriting criteria embedded directly into McGill's digital broking platform for near-instant deployment.
"This collaboration has the potential to disrupt the dynamics of the subscription market," said Steve McGill, CEO of McGill & Partners. "It strengthens the value proposition of leading underwriters in the market and redefines the way capacity is positioned in the best interests of our clients. This moves beyond incremental change, and repositions the way the market operates in the future."
How Does the Deal Work?
In the subscription market, large commercial risks are typically shared among multiple insurers. A lead underwriter prices the risk and sets terms, while follow underwriters decide whether to take a share of the placement at those conditions.
Completing a placement has traditionally been slow and labor-intensive, but with the risk already assessed and priced by a lead, algorithms can increasingly handle the following decision autonomously.
That is precisely what the company says the collaboration is designed to do. AIG conducted a detailed analysis of McGill's specialty portfolio, validating its quality and establishing underwriting criteria that can now be applied in real time through McGill's digital platform.
To get there, AIG worked with Palantir to build an ontology of McGill's entire portfolio using Palantir's Foundry platform. In practice, an ontology translates raw data into a structured map of how entities, risks, and relationships connect, allowing AI systems to reason about them rather than simply retrieve them.
The system generates near real-time insights on exposure levels, limit deployment, modelled risk outputs, and loss information, giving AIG continuous visibility into how its capacity is being deployed.
"The rapid evolution of AI and large language models is reshaping risk analytics, giving us the ability to continuously learn from McGill and Partners' portfolio and deploy capacity with greater insight, discipline and speed," said Peter Zaffino, Chairman and CEO of AIG. "By using McGill and Partners' robust data ingestion capabilities along with Palantir's Foundry platform, we are able to evaluate their portfolio to align with our risk appetite, and over time, we see significant opportunity to deliver greater efficiency to the subscription market while giving clients easier access to high-quality insurance solutions."
Why McGill’s Data Made This Possible
The deal did not happen by accident. It happened because McGill & Partners had the data to support it. Founded in 2019 with a deliberately digital-first approach, McGill built its broking platform from scratch. Structured, clean, and designed for exactly the kind of deep data analysis this collaboration required.
Without that foundation, AIG could not have built the ontology, run the portfolio analysis, or established the AI framework needed to manage capacity deployment on an ongoing basis.
This is not McGill's first AI-powered capacity deal. In December 2025, the broker launched a similar arrangement with AEGIS London using its proprietary Underscore platform for automated quoting and binding.
An earlier deal with AXA XL followed the same model. The pattern is consistent, and it points to something important. In the AI era, a broker's data infrastructure is becoming as commercially valuable as its market relationships.
This is also not AIG's first deployment of Palantir's technology in this context. In December 2025, AIG partnered with Amwins and Blackstone to launch Lloyd's Syndicate 2479, using Palantir's Foundry and multiple large language model agents to analyze more than four million industry data points across a delegated authority portfolio. The McGill collaboration extends that architecture into a new context, and at significantly greater scale.
For the subscription market, the implications are meaningful. Pre-secured capacity that can be deployed in near real time, governed by AI and continuously refined by live portfolio data, represents a fundamentally different model from the one the market has operated on for decades.