A consumer is asking ChatGPT which home insurance to buy, where to get the cheapest car insurance... right now. Not next year. Not when the technology matures. Now.
Every other distribution channel in insurance history gave carriers time to prepare. Aggregators arrived and carriers negotiated their position over years. Direct digital came along and carriers built capability at their own pace. Both channels sent an invitation.
This one did not. There is no account manager from OpenAI calling to discuss panel fees. The channel is forming around carriers, not being offered to them. That is a first.
In February 2026, OpenAI approved the first insurer-built app on ChatGPT, allowing a digital insurer to quote home insurance inside a conversation without the customer leaving the interface. Similar integration frameworks are already live on Anthropic's Claude. Google's Gemini is expected to follow.
The full purchase journey is not yet inside the chat. Right now the experience collects the customer's details conversationally, returns an indicative price, and redirects them to the insurer's website to complete the quote. That redirect is the part worth paying attention to.
42% of UK consumers say they are comfortable receiving an insurance quote from an AI chatbot, according to GlobalData's 2025 UK Insurance Consumer Survey. That number will not go down.
LLMs are already directing potential customers straight to insurer websites, bypassing the comparison table entirely. The battle for that traffic is already happening. If a consumer asks ChatGPT for the best home insurance and your product does not feature in the response, the customer is gone before they ever reach a price comparison site or your direct channel.
Before any carrier builds a formal integration, something else is already happening. LLMs are forming a view about insurance products based on whatever they can find. Product pages, policy summaries, coverage explanations, claims reviews across the web.
Most insurance product content was not written for this. It was written for a human scanning a comparison table, or a compliance team checking policy wording, or a customer service agent reading from a script. Dense policy language, coverage buried behind a quote journey, exclusions written in legal prose. A machine parsing that content for a recommendation cannot extract a clear picture of what the product covers, who it suits, or why it is worth choosing.
A carrier with plain, accessible product descriptions will surface better. Not because of a paid listing or a commercial relationship. Because the content is readable to the system forming the recommendation.
This is the new SEO. Google rewarded carriers whose websites were structured clearly and updated regularly. LLMs reward carriers whose product information is clear, specific, and honest about what they offer. Structured product data, plain language coverage descriptions, and schema markup that LLMs can parse cleanly are the technical levers most insurance marketing teams have never been asked to think about. The carriers that do will have a visibility advantage before they have built a single API integration.
Content is the first step. Infrastructure is the second. To be quoted accurately in real time inside an LLM interface, a carrier needs systems that can serve a quote to an external platform cleanly and quickly. Legacy policy admin systems built around web form journeys were not designed for this. Carriers with modern, accessible back-end infrastructure will be able to participate in conversations that others will not even know are happening.
MoneySuperMarket's stock fell 8.9% in the five days after the first insurer app went live on ChatGPT. The market called it a threat.
The picture is more interesting than that. 67% of UK adults used a price comparison site for insurance in the past year. Compare the Market and Go.Compare posted revenue growth of 21% and 28% respectively in 2025. The aggregator model is not broken. But it is being asked a question it has not fully answered yet.
The aggregators that engage with this shift could come out stronger, not weaker. They have things an LLM does not. Consumer trust built over two decades of UK advertising. Verified, real-time pricing data from regulated carriers. FCA authorisation and the commercial relationships that underpin it. An LLM scraping product pages is working from imperfect, static information. A price comparison platform feeding structured, verified, real-time data directly into an LLM conversation is providing something far more useful. The aggregator that becomes the trusted data layer inside ChatGPT rather than a destination website is not disrupted by this shift.
The rise of aggregators in the early 2000s changed the distribution landscape faster than most carriers expected. Some moved early and built real advantage. Others adapted later at greater cost.
The carriers watching this most closely are doing exactly that: watching. That is a rational position given regulatory uncertainty, legacy technology, and an industry that has learned to let others test new channels first. It is also the same posture that left some carriers behind when aggregators scaled.
For smaller carriers without the resource to lead on every new initiative, this distinction matters more than it does for the large players. The groundwork for participating in this channel does not require a transformation programme. It requires doing specific things well before the channel demands them, and building those into an existing distribution strategy rather than treating this as a separate initiative. The carriers that do it now do it once, at relatively low cost. The ones that wait will do it under competitive pressure.
Start with product content. Audit how your product pages, policy summaries, and coverage descriptions read to a machine parsing them for a recommendation. If an LLM cannot clearly extract what you cover, what you exclude, and why your product is worth choosing, the answer when someone asks ChatGPT for the best home insurance will not include you. This is not a technology project. It is a content and data problem that most marketing teams have not been asked to solve yet.
API readiness is the next question. If your systems cannot serve a real-time quote to an external platform cleanly, that is a distribution constraint as much as a technology problem. The carriers with accessible infrastructure will be able to participate in new channels at low marginal cost. The ones that have not done this work will face a rebuild at the moment when commercial pressure is highest.
Engage with the regulator before it engages with you. Where a carrier's own product documentation and communications are what an LLM draws on to form a recommendation, the accuracy of that information becomes a Consumer Duty question. A recommendation built on poorly structured product content could miss key exclusions, suggest cover that does not suit the customer's circumstances, or present details that are simply inaccurate. Carriers that take this seriously now, and engage with the FCA through its AI Lab and Sandbox, will have more influence over how the regulatory expectations around this channel are shaped. The ones that wait will find those expectations already written when they arrive.
Want to chat to us more about how AI is affecting the insurance world and what you can do about it? Get in touch.
[1] GlobalData, 2025 UK Insurance Consumer Survey, cited in Life Insurance International, February 2026.
[2] Reinsurance News, OpenAI approves insurer-built AI app on ChatGPT, February 2026. https://www.reinsurancene.ws/openai-approves-first-insurer-built-ai-app-on-chatgpt/
[3] Mintel, UK Price Comparison Sites in Financial Services Market Report, 2026.
[4] FCA, Regulatory Priorities for Insurance, February 2026. https://www.fca.org.uk/publication/corporate/regulatory-priorities-insurance.pdf