Takeaways:
- 2026 will see more insurers using agentic AI in their operations
- Agentic AI is well-suited to claims and underwriting workflows
Agentic AI became a buzzword for insurtech by the start of 2025, and experts expect it to catch on even more in 2026.
"Agentic AI is definitely showing a lot of promise. We are very, very positive about the technology, and we are quite excited about how we can leverage this technology," said Ashok Krishnan, chief innovation, data and analytics officer at AXA XL.
"With all of the focus that agentic is getting, and all the carriers we're talking to, in pilot or proof of concept with agentic, I think the percentage will be a little higher by the time we do this survey again in Q1 of 2026," said Keith Raymond, principal analyst at Celent, speaking in an October 22 webcast the consulting firm co-hosted with GlobalData, an AI and technology company.
Agentic AI can be applied to a variety of insurance functions, including claims adjudication, underwriting and evaluating an insurer's financials, according to Raymond.
- Claims adjudication: "Traditional machine learning can provide a score, but doesn't take action on the result," he said. "Agentic can manage the entire workflow. It's goal oriented. It analyzes a claim, decides next steps, retrieves policy data, compares it to external data like weather reports and vehicle databases. It generates a settlement recommendation or denial, and it can trigger the payment – all autonomously."
- Underwriting: Since AI agents can be constructed for numerous tasks, such as claims processing, evaluating first notices of loss (FNOL) and ensuring documents are complete, agents for underwriting, Raymond said, can be used "for medical questionnaire analysis, health history, diagnostic validation, income and occupation verification, and risk assessment."
- Operational Key Performance Indicators (KPIs): AI agents can examine financial KPIs, bottom lines and returns on investments, to justify additional funding for certain operations, according to Raymond. "We'll see a big uptick in the financial KPIs, and then ultimately the external KPIs for consumer interaction after that," he said. "A lot of organizations are still internal with their use of both Gen AI and agentic AI. Once the understanding of the technology is there, once the trust in the technology is there, then you'll see it more opened up into external-facing activities as well."
AXA's Krishnan expects to see an evolution in how insurers think about agentic AI and the technology itself. "More of the discussion is focused around agentics or AI agents within the construct of generative AI or LLMs, but it's a fairly narrow interpretation when you consider the overall AI ecosystem," he said. "It's very much early days."
