The Future of Insurance - Agents & AI: Same Channel, Different People

Episode Detail

This episode is a discussion of the thought leadership paper Bryan Falchuk published in September 2026 on a pressing an important question the industry is facing today – will AI spell the end of insurance agents and brokers?

AI is not going to kill the insurance agent channel. That's not really in dispute, and it was never the right question. In this solo episode, Bryan Falchuk argues that the public debate over AI and agents has been fought almost entirely on the wrong terrain — demand — when the real disruption is on the supply side. Buyer demand for a trusted expert on a complicated, high-stakes decision is as durable as ever, but that says nothing about how many people will be doing that work five years from now, or how much of it AI will be doing on their behalf. Bryan traces the argument through a real disagreement with fellow industry voice Matteo Carbone, a landscaping-business example of what agentic AI can catch that no human ever sees, and the two mechanisms he thinks actually create opportunity inside this disruption: Empowered Expertise and Market Expansion.

The channel isn't disappearing. How it's staffed, and by whom, is about to change faster than most of the industry is planning for.

Show Notes:

The Real Question: Demand vs. Supply

Demand isn't the debate: Bryan agrees the agent channel survives — buyer demand for a trusted expert on a complex, infrequent, high-stakes decision isn't eroding.

The real question is who or what does the work inside the channel five years from now.

A channel can hold 100% of its market share and still be staffed by a fraction of today's headcount, partly by AI standing in an agent's seat.

Where the Reframe Came From: The Matteo Carbone Debate

The idea grew out of a real disagreement with Matteo Carbone, Founder & Director of IoT Insurance Observatory, who argues buyers won't shop and transact alone for something this complicated. Bryan agrees.

Bryan's pushback: demand-side evidence, like low churn, doesn't say anything about how much of the work AI ends up doing.