AIG AI: From digital twins to underwriting agents
AIG used its fourth quarter 2025 earnings call to signal that artificial intelligence is moving from experimentation to core infrastructure, with generative AI now embedded across underwriting, portfolio management, and operations, alongside a new push to orchestrate AI agents at scale.
The company said it has made “significant progress embedding gen AI across our core underwriting and claims processes,” with the technology expanding across multiple lines of business. Its Underwriting by AIG Assist tool has been rolled out to seven additional lines, including Lexington, and is scheduled for broader deployment across North America, the U.K., and EMEA in 2026.
Early results point to measurable operational impact. Lexington recorded a 26% increase in submission counts year over year, while its middle market property business improved its submit-to-bind ratio by 35%. AIG has already reached more than 370,000 submissions, putting it on track toward its goal of 500,000 by 2030. Management said the objective is clear: “We believe our use of gen AI gives us a strong advantage going forward in this dynamic market.”
Beyond underwriting productivity, AIG is using generative AI to support strategic transactions. During the integration of renewal rights from Everest’s global retail portfolio, the insurer used gen AI to analyze the book and prioritize accounts “in a fraction of the time.” The company built what it described as an Everest ontology, effectively a digital twin of the portfolio, enabling teams to evaluate limits, pricing, attachment points, and conversion strategies. The same capabilities, delivered through AIG Assist, helped accelerate renewals during the transition.
AI is also shaping capital deployment. In the formation of Syndicate 2479, AIG partnered with Palantir to use large language models to match Amwins program data with the syndicate’s risk appetite. According to management, the approach helps identify aligned opportunities while also informing portfolio construction through advanced analytics. The transaction marked the first time AIG deployed its generative AI capabilities in a third party capital structure, and the company indicated it has a strong pipeline of similar opportunities.
CEO Peter Zaffino said rapid improvements in large language models have accelerated the company’s ambitions. “What we had outlined in March was aspirational and six to twelve months later, we see the capabilities are much greater.”
Looking ahead, AIG’s most forward leaning initiative for 2026 is the development of an orchestration layer designed to coordinate multiple AI agents across the enterprise. The company described these agents as operational companions with defined roles, including knowledge assistants that provide real time information, adviser agents that generate insights based on historical cases, and critic agents that challenge recommendations and decisions. The orchestration layer will determine when agents are activated, how they share information, and when human oversight is required.
Zaffino emphasized that the focus is shifting from isolated use cases to enterprise scale. “The implementation of single agents throughout organizations is real, but orchestrating that in an orderly way of being able to get that at scale is what we’re going to focus on in 2026.” The effort will span front office, mid office, and back office functions, including work with outsourcing partner Accenture, with AIG sharing in both the design and the efficiency gains.
