How 157 insurance executives use and govern agentic AI as it moves into claims, underwriting, and servicing.

Establishing Accountability Before Scaling Autonomy

Insurers are deploying agentic AI directly where it touches the customer. 92% use agentic AI in customer service, compared to 69% across industries, and 64% in sales and marketing, compared to 36%. As agents expand into claims, underwriting, and servicing where decisions affect coverage, price, and payouts, every decision has to be one the insurer can reconstruct and explain to the policyholder and to the regulator. 

OutSystems and KPMG partnered with CIO Dive to survey 157 insurance executives. The report shows where adoption is moving fastest, and what insurers still need to address as agents get closer to the policyholder.

Key Takeaways from the Report:

• Automate Inside the "Yes Zone": Let agents act where clear, documented rules define the steps, and route everything else to an adjuster or underwriter.
• Ground Actions in Governed Context: Give agents the policy wording, coverage limits, and case facts they need, while restricting sensitive data.
• Measure the Whole Outcome: Look beyond task speed to track operational gains like cost per file and cycle time alongside policyholder metrics like quote-to-issue and renewal rates.
• Expand Autonomy Only with Evidence: Compare agent-supported processes against your pre-agentic baseline before extending automation into higher-risk areas.