Salesforce lost its head. Insurance is next

I think the future of insurance distribution is headless. Headless quote and service engines, headless payment and credit facilities, headless claims management and fulfilment. Key elements of the insurance proposition exposed directly via MCP and APIs, so any AI agent can query them in real time. No website. No portal. No screen.

Salesforce is already there. This year it did something remarkable for a company its size - it removed its own head. Its Headless 360 platform exposes the entire CRM as MCP tools and APIs, so staff and customers can use Salesforce from inside Claude, ChatGPT, Slack or Gemini without ever opening it. With Anthropic, it has gone further still, launching Claudeforce to put the CRM directly inside Claude. As Marc Benioff, Salesforce CEO, put it: “the UI is the AI.”

Salesforce chose to disrupt itself before someone else did. Insurance should pay attention, because so much of our distribution, especially in personal lines, is built on owning the screen.

If insurance goes the same way, several models come under pressure.

The PCW model. Comparison sites win by owning the place where the customer journey starts. If the journey starts in Claude or ChatGPT instead, the ranked list and the CPA lose their purpose.

The software house model. Much of our market runs on platforms whose value lies in the quote journey, the front end and the insurer integrations. When insurers expose their engines directly via MCP, a lot of that layer isn’t technically needed. At the very least, it isn’t needed in the same way, or at the same price.

The premium finance model. Customers no longer need to accept the retailer’s preferred instalment plan, some at eyebrow-raising APRs. Other payment providers could compete for customers at radically different price points.

The traditional broker model. If insurers no longer need to sit on software house panels, and quote traffic shifts away from PCWs, brokers built on that model will struggle. But brokers who reinvent themselves could come out on top.

So, if those layers fall away, who sits between the customer and the insurer?

Probably not the AI platforms. I’d be very surprised if OpenAI, Google or Anthropic wanted to be regulated by the FCA. Insurance is hard. Claims, complaints, vulnerability, Consumer Duty, the FOS. Just look at Amazon, which launched a UK insurance store and closed it about 15 months later.

Nor individual insurers on their own. esure has just launched car insurance quotes inside ChatGPT, and credit to them for moving early. But customers won’t install 100 insurer plugins. They’ll want whole-of-market coverage in one place; from someone they can hold responsible.

That’s the gap for a regulated orchestrator. One that sits between the customer’s AI and the market, connects to every insurer’s endpoint, and takes accountability for the outcome.

Then there’s the customer proposition. MCP-enabled search inside today’s chat interfaces is step one. The bigger prize is agentic. A team of AI agents working for each customer, each accountable for a key part of the process. One understands needs, one searches the market, one checks cover and value, one handles payments, one supports claims, and one watches every renewal.

That’s not a better comparison site. It’s a different way of retailing insurance.

The FCA has a hard balance to strike here. Customers need protecting, and guardrails such as accountability, fair value, explainability, and good outcomes for vulnerable customers are critical. But this could be the biggest disruption our market has ever seen, and innovation needs room to happen.

Regulation may yet be what protects the incumbents.

The question is whether that’s the right outcome for customers.

Which of these models adapts, and which gets left behind?