Is One AIG Enough in the Age of AI?
American International Group (AIG) is taking a major step forward in artificial intelligence by building a digital twin of its organization — a virtual mirror of the business powered by generative AI, according to CEO Peter Zaffino.
Speaking during a recent earnings call, Zaffino explained that the digital twin will serve as a comprehensive model of AIG’s core operations, including its data, processes, business rules, and the relationships across departments and business units.
“Our goal is to create a digital version of our enterprise that captures how everything connects and operates,” Zaffino said. “This is a key part of our AI strategy, and we’ve been developing our AIG ontology from the very beginning.”
He added, “Expect to hear more in upcoming quarters as we continue to scale this initiative.”
Understanding Ontology in Business and AI
While “ontology” has philosophical roots — referring to the nature of existence — in business and AI, it describes a framework for understanding and structuring what an organization needs, and how those needs interact.
In the AI world, ontology gives structure to data, allowing machines to not only process information but also apply logic, make decisions, and improve operations.
Zaffino emphasized that AIG’s ontology will power technical reviews, support claims personnel with better tools and insights, enhance decision-making, and strengthen client service.
He also noted that this system will maintain a traceable log of actions taken across the organization, supporting business logic and providing transparency for audits.
Powered by Partnerships and AI Infrastructure
Zaffino highlighted AIG’s collaborations with Palantir, Anthropic, and Amazon Web Services (AWS) in building its AI infrastructure.
Palantir, in a blog post, describes ontology as essential for enterprise-level decision-making, helping AI understand not just data, but the context and reasoning behind actions.
“Traditional data systems don’t capture why decisions are made or what results they produce,” wrote Palantir’s chief architect. “They stay disconnected from real operations. Today’s enterprises need a decision-focused software architecture.”
By leveraging AI through a digital twin, AIG hopes to model decisions, simulate outcomes, and make more informed, proactive choices at scale.
Real-World Applications Already Underway
AIG isn’t just planning — it’s already implementing AI in targeted areas.
Zaffino shared that AIG began using GenAI to assist underwriters for nonprofit business lines late in the first quarter. “The early signs are very positive,” he said.
Next, the AI will roll out to middle-market accounts at Lexington Insurance in the third quarter, with a broader expansion across North America commercial lines planned for 2026.
This technology was previewed earlier at AIG’s Investor Day, where the company discussed its ability to amplify underwriter capacity through AI — effectively multiplying productivity. (See: AIG: Turning One Human Underwriter Into Five, ‘Turbocharging’ E&S)
AIG is also using GenAI to enhance its claims operations. In early tests, AI reduced notice-of-loss processing times from days to just hours.
“We’re training large language models to extract and organize key insights automatically,” Zaffino explained. “That way, claims adjusters can make quicker, more informed decisions — helping us deliver on our promise to clients when they need it most.”