AIG
AIG's AI assistant for underwriters speeds quoting and lifts binding in Lexington middle market property
AIG launched Underwriting by AIG Assist in 2025 to help underwriters review insurance submissions with more and better information in less time, and has extended it to 8 lines of business. On its Q1 2026 earnings call, AIG's CEO reported a 55% cut in time to quote in Lexington middle market property.
Key figures
Copied exactly from the sources. Copy any figure with its citation.
30%
Improvement in quoting more submissions with AIG Assist, Lexington middle market property (per the CEO)1
55%
Reduction in underwriters' time to quote, Lexington middle market property (per the CEO)1
approximately 40%
Increase in binding of submissions, Lexington middle market property (per the CEO)1
What it took
Cost, team size and timeline were not disclosed in the sources.
The problem
Commercial insurers receive large volumes of submissions, the files brokers send to ask for cover. Underwriters must read them, pull out the facts and decide whether to quote. AIG wanted its underwriters to see more and better information in a fraction of the time, so they could quote and bind more of the business it wants to grow.
What they built
In 2025 AIG launched Underwriting by AIG Assist, a generative AI tool that helps underwriters review submissions. After the launch, AIG began deploying it across 8 lines of business. The sources do not name the model.
How they did it
AIG targeted Lexington middle market property, a segment it wants to grow, as the area to show results. The CEO says AIG sees an effect on underwriting metrics and on data quality. The CEO also describes a later phase in which AI agents review submissions and recommend observations so underwriters can decide with more information, building on AIG Assist.
Results
For Lexington middle market property, the CEO said AIG Assist helped deliver a 30% improvement on quoting more submissions. It cut underwriters' time to quote by 55% and raised binding of submissions by approximately 40%. These are the company's own statements on its first-quarter 2026 earnings call.
Lessons for your project
Prove the tool in one growth segment before you spread it. AIG reports its numbers for one targeted area, Lexington middle market property, not for every line.
Pick measures that link to the business goal. AIG tracks how many submissions get quoted, how long a quote takes and how many are bound, not only time saved.
Keep the underwriter in charge. AIG says its AI is meant to give underwriters more and higher-quality information, with underwriters still making the decision.
Build on a working tool. AIG says its next phase of AI agents builds on the early results of AIG Assist.
Caveats
- All figures are management statements on an earnings call. The sources do not say how the results were measured or compared, or over what period.
- The model behind AIG Assist is not named. The CEO names Anthropic and Palantir for the follow-on multi-agent phase, which is separate from AIG Assist, so no model is attributed here.
- Only the year of launch (2025) is stated in the call, so the deployed date shows the year only.
- The second source, Reinsurance News, reports the same call and repeats the same figures. It does not independently check them.
Sources
- AIG (AIG) Q1 2026 Earnings Call Transcript
The Motley Fool (transcript of AIG's Q1 2026 earnings call) · 2026-05-01 · Earnings call · fool.com
- AI advancing faster than expected as AIG builds multi-agentic solution: CEO Zaffino
Reinsurance News · 2026-05-01 · News · reinsurancene.ws
Summarized in our own words from the sources above. Spotted an error? See corrections.