Fear, not technology, is holding insurers back on AI: Five Sigma CEO

Ahead of the Agentic and Generative AI for Insurance conference in New York on November 18–19, 2026, Intelligent Insurer spoke to Oded Barak, CEO and co-founder of Five Sigma, who will join a panel discussion on scaling AI in insurance, from experimentation to enterprise-wide adoption. He shares his views on what is holding insurers back, where AI can deliver the greatest value and what it takes to move beyond pilot projects.

What separates insurers scaling AI from those still experimenting?

Mostly its indecision driven by fear, not the technology. Pick your biggest pain point, set a success threshold, and once the pilot clears it, just do it.

Many insurers tend to treat AI as an IT project to be de-risked. Others, and our partners, that scaled it, and other industries like banking and e-commerce, treated it as a business transformation with clear ownership and measurable outcomes.

It’s the same thing that separates people who reach their potential from those who don’t: not letting fear stop them from acting.

What infrastructure and safeguards are needed to scale AI?

Two things: solid integrations and APIs into core systems, and real change management so the workforce actually adopts it.

Set clear guardrails up front and give AI more authority as confidence grows, but don’t let governance become your excuse to wait.

Where is AI adding the most value, and how will agentic AI change roles?

In claims, the biggest value comes from automating the entire workflow end to end, not point solutions for documents or bill review.

Professionals will go from doing the work to supervising it, making sure workflows run correctly and stepping in on edge cases and escalations.

What do insurers get wrong about moving AI into production?

The biggest misconception is that a POC has to be perfect before you go live. Production is where AI actually learns and improves. For example, a carrier that rolls out FNOL automation with a human reviewing every claim, then removes the review step as accuracy proves out, gets to scale much faster than one stuck tuning a pilot for 18 months.

What do you hope to discuss with fellow panellists at the Agentic and Generative AI for Insurance conference in New York?

Hearing about their real pain points and sharing how we can help solve them.

Did you get value from this story?  Sign up to our free daily newsletters and get stories like this sent straight to your inbox.