Liberty Mutual says AI must be embedded across insurance operations

Liberty Mutual was rolling out a secure internal GPT platform to 74% of eligible employees, putting generative AI inside everyday work rather than parking it in an innovation lab. The move fits a longer pattern at the Boston insurer, where AI has already been pushed into underwriting, cloud operations and employee tools.

That operating model was visible as early as May 26, 2020, when Liberty Mutual announced a commercial underwriting AI initiative with Groundspeed. The company said then, “We use AI to enhance risk selection, speed underwriting and add even greater value to our broker partners by identifying growth opportunities.” Liberty Mutual’s own materials say the system helps extract data from unstructured submission documents so underwriters can assess risk more quickly and accurately, a practical use case that sits directly inside the underwriting desk rather than alongside it.

The scale of the technology stack makes that stance more credible. An AWS case study said Liberty Mutual processed 100 million transactions a month in its FCS system, brought computing costs down to $60 per million transactions, and deployed more than 3,500 serverless patterns in one year using AWS CDK. The same case study said the company cut application build time from one year to three months and ran 50% of its workloads on AWS in 2020. For a P&C carrier, those are not abstract cloud milestones; they are the plumbing that determines whether AI can be embedded into core workflows or stuck in isolated pilots.

Liberty Mutual also put real money behind the strategy. The company announced a $25 million, five-year collaboration with MIT to support AI research in computer vision, language understanding and decision-making. By March 19, 2025, LibertyGPT was being used by more than 10,000 employees, and a January 19, 2026 source said Liberty Mutual was rolling out a secure internal GPT platform with direct access for 74% of eligible employees. Monica Caldas, the company’s global CIO, has said AI is “a core part of how we run the business,” and has framed the strategy as more than technology, tying it to culture, operations and human potential.

Caldas has also discussed AI adoption and risk management at Liberty Mutual in MIT Sloan Management Review, underscoring the governance side of the program. That is the real dividing line for P&C software buyers: AI only matters when data, workflow design, model oversight and business ownership are built into underwriting, claims and service systems, not bolted on after the fact.