AIG Implements Generative AI to Enhance Underwriting and Claims

American International Group (AIG) has announced accelerated benefits from the adoption of generative artificial intelligence (AI) technologies across its core insurance operations. According to disclosures during its recent Investor Day, the firm cites measurable gains in processing capacity, cost reduction, and workflow redesign.

The deployment of generative AI, a type of model capable of creating new content or summarising complex data, is now central to AIG’s underwriting and claims functions. CEO Peter Zaffino acknowledged that the company’s initial projections were aspirational, but recent internal results suggest a more rapid transformation than anticipated.

“We’re seeing a massive change in our ability to process a submission flow… without additional human capital resources. That has been the biggest surprise,” Zaffino said in a recent earnings call. The firm reports that generative AI has increased submission processing capacity and is now deeply embedded across most commercial business lines using an internal tool called AIG Assist.

Lexington Insurance, AIG’s excess and surplus lines unit, has already surpassed 370,000 processed submissions for 2025, approaching a target of 500,000 by 2030. Generative models are now used to extract and summarise incoming data, supported by an orchestration layer within AIG’s technology stack. This orchestration layer coordinates multiple AI agents, which act as digital "companions" to staff by providing real-time insights, drawing on historical cases, and challenging underwriting decisions to improve quality.

Agents managed via orchestration not only accelerate intake and risk assessment but also compress the entire workflow from initial submission to claims handling. AIG reports managing and analysing data “at a fraction of the time” previously required, streamlining processes that were once repetitive and time-consuming.

AIG has put its generative AI system to use in complex portfolio integration projects. For example, during the integration of Everest’s retail commercial business, AI models helped build and align ontologies—data frameworks that map portfolio attributes—enabling prioritisation for renewal and efficient merging of account information. The CEO noted that such ontological alignments are technically complex and often underestimated in cost and effort.

The company extended its AI-driven approach to new ventures, including the launch of Lloyd’s Syndicate 2479 in partnership with Amwins and Blackstone. By working with technology provider Palantir, AIG used large language models (LLMs—a term for advanced generative AI trained on extensive text) to check alignment between Amwins’ portfolio and the syndicate's risk criteria. Management indicated that the company is building a “strong pipeline of SPV [special purpose vehicle] opportunities.”

The case demonstrates the tangible economic impact of embedding generative AI in underwriting, claims, and portfolio management. Measurable improvements in cycle times and capacity highlight the technology’s ability to drive business transformation, particularly for large-scale insurers handling diverse and complex portfolios.

AIG’s experience may provide a benchmark for other insurance organisations seeking to integrate generative models and orchestration frameworks into mission-critical workflows.

artificialintelligence-news.com