ChatGPT is now a distribution channel. Are you in it?

We're at the start of a major shift in how people find and interact with insurance products.

Over the past eighteen months, hundreds of millions of people have started using AI assistants like ChatGPT not just to write emails or summarise documents, but to make decisions. Topics like "what health insurance should I choose?", "do I need life insurance?", "is my home underinsured?" are becoming commonplace. They speak to the opening moments of a buying journey, and they're increasingly happening inside AI agents.

For anyone retailing insurance products, this represents something genuinely new: a distribution surface that didn't exist three years ago, growing faster than almost any channel in the history of financial services, and one where - right now, at least - very few insurers have a deliberate presence.

Moving from search to context

For decades now, Google has been the place where insurance intent surfaces. Someone types "cheap car insurance" or "life insurance for over 50s", and a complex algorithm - massaged and optimised for by experts in SEO - presents back the most relevant responses, by some measures.

That funnel's changing. ChatGPT handles over a billion queries per week, and a growing proportion of those queries are research-led, advice-seeking, and purchase-adjacent. Unlike a search engine, which returns a list of options and pushes the user to decide, a conversational AI synthesises, makes a determination about the user's intent, and tries its best to satisfy it. When a user asks "what's the best income protection insurance for a self-employed person?", they're not looking for pages of blue links - they're looking for a convincing answer. Right now, AI can't safely provide that for regulated products, but the intent's there.

The insurance-specific data tells the same story, perhaps to an even starker extent. According to Insurify, 42% of drivers have already used an AI assistant to help shop for car insurance - and in some segments, that figure rises to 60%. 86% of Americans say they would trust AI to assist them in buying car insurance in some form, whether comparing quotes, explaining coverage, or guiding them to a decision.

The models enabling this shift are only getting more sophisticated at serving those answers. And unlike on Google, the answer doesn't - yet - have to be bought. It can be earned.

What actually makes this change different

By our count, there are three things that make AI-first insurance fundamentally different to its predecessors.

It's consultative by default. Traditional digital channels are transactional: click, compare, buy. AI is more discursive than that - it asks clarifying questions, surfaces trade-offs, and builds the kind of context that can currently only be achieved in conversation with a professional. This means it has the potential to surface products in a far more user-friendly way - provided it has the right tools at its disposal to do that safely.

It compresses the journey. At the same time as providing more information, AI offers the potential to simplify the existing process dramatically. The average insurance purchase currently involves a few different touchpoints across days or weeks - a Google search, a comparison site, a return visit, a quote form. AI can collapse that into a single conversation. Carriers that are able to be part of that conversation - surfacing products, answering underwriting questions, quoting in real time - gain a meaningful advantage.

It reaches underserved segments. A significant portion of AI users are younger, digitally native, and don't come with baked-in learned behaviour of existing channels. They value simplicity and openness, and are less likely to engage over the phone or by email. For carriers struggling to acquire Gen Z and millennial customers at a reasonable cost, this is a really promising avenue.

So what's missing?

Given how new this model is, it's not surprising that not everyone is completely ready for it. Most insurers aren't yet discoverable inside AI systems - their products aren't structured for AI retrieval, their APIs aren't connected to the surfaces where purchase intent is forming, and their underwriting logic and product information aren't accessible to the models that are, right now, answering questions their potential customers are asking.

This is partly a marketing problem, but it's mostly an infrastructure problem - and it requires a different kind of solution than rolling out another microservice or app. It requires connecting product logic, underwriting rules, and quoting capability to the interfaces where modern consumers are spending their attention. That's not an easy, or static, problem to solve.

That said, tackling this challenge unlocks more than volume now and strategic positioning going forward. It also establishes data flywheels, brand familiarity, and distribution relationships that compound over time - much in the same way that early movers on comparison aggregators locked in vantages that took rivals years to erode.

What good looks like

We'll talk more about this in a future post, but there are green shoots to look to in other sectors, like Claude's Financial Services tooling. You can see a lot of the same threads of data availability, connectivity, and agentic flexibility coming to bear - and the same's starting to hit insurance.

One thing to ponder

The question isn't so much whether AI will become a meaningful insurance distribution channel; it's really a question of when - at scale. Most insurance products, as they're constructed today, are tightly coupled to existing distribution channels. If you were to roll out to AI channels tomorrow, how would you do that - and where do you see the opportunities for improvements in user experience?

The "right" answer to that question won't look the same in all lines of business, nor every channel - but it's the insurers and brokers who start thinking seriously about those questions now that will set themselves up to win in the next decade.