Shopping and switching is fundamentally deflationary to the overall insurance market. If brokers are better at their job, then the set of people who are paying over the odds for an insurance policy they could buy more cheaply elsewhere will decrease. The combined ratio of the industry as whole is kept down by consumers who are not making purely rational and profit-maximising decisions; their stickiness creates surplus that funds everyone else. Much has been made of the potential for AI to disrupt this happy equilibrium; Matt Harris from Bain Capital Ventures wrote as early as June 2023 (!) that “Generative AI Is the Attila the Hun That’s Going to End the Banking System as We Know It”; the Scourge of God is now coming for insurers.
We believe that agentic shopping and AI assistants represent the most important change to hit personal lines this century.
AI assistants are incredibly useful but currently unprofitable. Assistants cost anything from tens to hundreds of dollars a month to run, even though roughly 95% of Americans don’t spend money on AI subscriptions. As Elon Musk tweeted on 1st November, “the cost of serving Grok Bot was unsustainably high relative to revenue.”
Agentic commerce promises to address this monetisation gap, by re-routing consumer spend and taking commission on it. There’s a long way to go to realize this dream: almost all American households spent zero dollars through AI agents in H1 2026. In general, the AI companies (e.g. Meta) are bullish, the merchants (Uber, Amazon, Walmart, Chase etc) are bearish, and payment companies are somewhere in the middle.
For instance, Mastercard’s CEO Michael Miebach said on Oct 29 that:
“It is still early days, and the numbers are not yet significant in the grand scheme. As the AI agentic assistants have emerged, agentic transactions are ramping up at impressive weekly growth rates.”
While Stripe won’t share more detailed numbers, Decoder reported that “agent transactions on [Stripe’s agentic payments product] Link grew 38x and volume 20x in September”.
We believe that home and auto insurance deserves particular focus, as one of the most interesting categories for agentic commerce. J.D. Power reports that Americans are already switching home and auto insurance policies at a rate of 2% and 4% per month respectively; we expect this number to increase as AI assistants start to drive shopping.
But the carriers are not yet seeing significant AI traffic. Tricia Griffith, the CEO of Progressive (America’s leading auto insurer, with almost 20% market share), said on November 3rd that:
“Over the last few weeks, we estimate we received agentic inbound interactions representing a rounding error relative to our overall volume. Activity we see today suggests much of this usage reflects experimentation rather than true shopping intent. While the usage seems limited today, we believe there is a high likelihood that agentic AI changes some aspect of consumer shopping behavior.”
Meanwhile, Progressive’s Chief Personal Lines Officer Lori Niederst said they are “preparing, assuming that agentic shopping has the potential to become an important acquisition channel for us.”
Michael Klein at Travelers said that “the actual volume of observable activity is pretty minor”, and Lemonade’s CEO Dan Schreiber said on Nov 5th that “we're not seeing it yet in the volumes”.
An important piece of context here is that online ad spend in home & auto insurance is really high right now. In a soft personal lines market, insurers are trying to buy market share through paid acquisition spend, which meant that MediaAlpha (NYSE: MAX), the ad exchange through which most online insurance ad spend is routed, hit a 1-year high after Q3 earnings were announced. Yet agentic shopping threatens spend on online direct-response ads; it’s not clear that an AI assistant will be at all influenced by the information that Progressive or GEICO can sell you an insurance policy. If AI assistants are going to be less swayed by brand familiarity than human shoppers, this is extremely bad news for large incumbents that have invested heavily in their brand.
As such, MediaAlpha CEO Steve Yi laid out a defensive position on Oct 28:
“No, we have not seen any impact of personal AI agent shopping on behalf of consumers… I think ultimately when it comes to quoting and binding, the carriers are going to retain control of those aspects of the consumer experience. First of all, consumers [sic] have invested tens if not hundreds of billions of dollars in building brand awareness and getting consumers to interact with them directly, be it one of their agents or through their website. The carriers are not going to relinquish that control easily or lightly by allowing a personal AI agent to dictate and control how a consumer is going to access their site and systems.”
This is the crucial point. Some carriers (but not all!) are unwilling to let AI assistants shop for insurance - even if that experience is the better consumer experience! Steve is correct that carriers have spent hundreds of billions of dollars on their brand; they’re unwilling to let an agent compare quotes in a purely rational and Procrustean manner.
On the other hand, there are signs that AI assistants are more likely to recommend the most popular brands; as David Pawlan shared, a paper published by Italian researchers on 5th November suggests that dominant brands in the top quintile are cited 1.4× to 2.8× more frequently than average brands. That’s great news for the carriers which spent hundreds of billions of dollars on branding!
Equally, it’s important to understand that showing real quotes requires insurers to buy data reports from vendors; each marginal quote costs them money. This, combined with the lost advertising mechanism, was why Insurify initially blocked Muse. A key question in agentic commerce remains: where and why does the system break if volume increases by a factor of 10, while conversion rates fall by an offsetting factor of 10? Insurers need a better answer to all this problem - and it’s something we’re actively addressing at Foresee.
In general, we see these developments as an unabashed positive for consumers. They will be better equipped than ever to find the right products (financial or otherwise) at the the best possible price. But incumbent vendors will need to adapt - and so will the intermediaries who enable them to transact with consumers.