This week, over $25 billion in insurance market cap evaporated in a single session because a Spanish MGA put home insurance quotes on ChatGPT.

Marsh fell 7%. Aon dropped 9%. Gallagher shed nearly 10%. The worst day for insurance brokerage stocks in decades.

People saying this is the end of the broker are wrong. I’ve built and deployed AI inside large carriers and brokers. I’m in the room. I see what these companies are building, what the technology actually does today, and where this is going.

The market has the winners and losers dangerously backwards.

What I’ve actually seen

Two years ago, AI couldn’t reliably read an Acord form. It would pull wrong fields, hallucinate coverage limits, and confuse named insureds. A year ago, it could handle simple personal lines if the format was clean. Today – not in a pilot, but in production – I watch AI process complex multi-line commercial submissions end-to-end. It extracts data from messy documents that would take a human team hours, flags coverage gaps, and routes complete packages to underwriters. In minutes.

The executives at the companies deploying this aren’t surprised by yesterday’s headlines. They’re shaking their heads. Because they know what this technology is actually doing for them and it’s the opposite of what the market thinks.

“We’ve heard this before”

Every InsurTech wave for fifteen years has predicted the death of the broker. Comparative raters. D2C platforms. Digital MGAs. And every time, the broker survived. Why? Because carriers don’t want to go direct, risk selection bias crushes D2C economics, and complex businesses genuinely need expert advice.

None of those structural realities changed yesterday.

But here’s the thing that makes this genuinely different from every previous wave, and it’s the most important thing I can say in this piece:

Every previous attempt at disruption was a product: a website, an app, a startup trying to build distribution from scratch. Those had to compete with incumbents who had decades of relationships. They mostly failed.

AI is not a product competing with you. It’s an infrastructure shift that makes your own business dramatically better.

That is the difference. And it changes everything.

The story the market is missing

Insurance was always supposed to be about two things: understanding risk and serving customers. 

But over decades, we built armies of people to handle the bureaucracy. Policy servicing, claims processing, submission intake, data rekeying. Carriers wanted to run sophisticated risk-selection operations. They ended up running back offices. Brokers wanted to advise clients. They ended up buried in paperwork.

Here’s something most people outside the industry don’t realize: many carriers today aren’t even reviewing 100% of the submissions they receive. Not because they don’t want to but because the operational burden makes it physically impossible with human staff alone.

AI strips the overhead. What’s left is the actual business. And it’s a dramatically better business.

Margins expand. Speed to market increases. And carriers and brokers can now profitably serve segments that were previously margin-negative. New products. New customers. New revenue that didn’t exist before.

This disproportionately benefits the largest players. They have the customer base to deploy across, the scale where efficiency gains compound, the capital to invest, and the data that makes these systems effective. Scale is the moat. AI deepens it.

The smartest leaders in the industry already know this. Several of the largest public and PE-backed insurers have deployed AI across core operations – not as experiments, but as CEO-level transformations. These are the same companies the market sold off yesterday.

My prediction: many of the companies that got hit hardest yesterday will be worth more 18 months from now than they were before the crash - not despite AI, but because of it.

Nobody buys a seven-figure policy from a chatbot

Let me be specific about where the threat is real and where it isn’t.

Personal lines and simple small commercial? Yes, they will face pressure from conversational AI. That’s the Tuio thesis, and it has merit.

But nobody trusts a language model to redesign their risk program after an acquisition. Nobody calls ChatGPT at 2am when their warehouse is on fire. Nobody asks an AI to sit across from their board and explain why their cyber liability limits need to triple.

Large commercial, specialty, E&S, complex risk programs — this work requires human judgment, relationship depth, and domain expertise that AI augments but does not replace. This is the work that drives the economics of every major brokerage. And with AI handling the operational burden, the people doing this work are about to have more time to do it better than ever before.

Across the value chain, the opportunity is enormous. Carriers reclaim margin that was going to outsourced operations. Wholesalers and MGAs deepen their specialty expertise as the market for specialized underwriting grows. The winners at every level will be the ones who know what only they can do, and use AI for everything else.

What the leaders are getting right

In my conversations with the executives moving fastest on this, a few patterns stand out.

They've made this a top-of-house priority. Not an IT project. A CEO-level transformation of how the business operates and delivers value.

They're reinvesting operational savings into their people and their customers, freeing up capacity so their brokers, underwriters, and service teams can spend time on the work that actually differentiates: deeper client relationships, more sophisticated risk consulting, faster and more responsive market access.

They're expanding into new segments. Lines and customer sizes that didn't make economic sense before are suddenly viable when the operational cost to serve them drops dramatically.

They are moving now. The window for first-mover advantage is closing. 

This is the moment

The actual work of understanding risk and advising clients - that work is not being replaced. It’s being unleashed. Every hour you spend on operational overhead is an hour AI is about to give back to you. So you can do the work that no chatbot will ever do.

This is not the beginning of the end for insurance. It’s the moment the industry sheds decades of operational debt and gets back to what it was always supposed to be about.

The companies that see AI as a threat are right to be scared. The ones who see it as leverage are about to dominate. The market just gave you a 10% discount on the winners. I know which ones I'd bet on. 

Jamie Cuffe

Founder, Pace