Atlantas.Ai A Kwick Blocks product

Blogs · July 2026

Can an AI agent handle a claim notification?

Yes. It captures the claim and passes it on, inside your permissions. The safe model is simple: capture the claim, never advise on it.

The short answer: yes. An AI agent can take the claim details, capture the evidence and pass the claim into your process, and that is one of the most useful things it will ever do for you. The safe operating model is simple: capture the claim, never advise on it.

Worth knowing, because it is easy to assume otherwise: taking a claim notification is not simply admin. Helping a customer notify a claim and put evidence behind it is itself a regulated activity under FCA rules when it is done for the policyholder. Your firm is already authorised for that work, so nothing changes except who does the typing. The agent operates inside your permissions and your process, not around them.

The distinction that matters

The FCA separates two things:

  • Merely giving pointers about how to fill in a claim form falls outside the regulated activity.
  • Notifying a claim and then providing evidence in support of it is "likely to amount to" assisting in the administration and performance of a contract of insurance, which is a regulated activity when done for the policyholder.

In practice, a broker taking a first notification of loss (the initial call when a customer tells you something has happened) is already authorised for this work. Nothing about an AI agent changes that. What changes is who does the typing, and your firm remains accountable for all of it.

So the useful question for a supplier is not "can your AI take a claim". It is: "where does the regulated line sit, and how does your agent stay on our side of it?" A supplier who can answer that clearly is one worth talking to.

Why you would want it doing claims at all

Because the claims experience is where brokers quietly lose people, and the failures are administrative rather than technical.

Which? surveyed 3,322 claimants and found:

  • 20% had to chase for a status update.
  • 21% had to repeat information or resend documents they had already provided.
  • 26% said their initial contact left them no clearer than before.
  • 17% reported delays in the insurer communicating a decision.

None of that needs an adviser. It needs someone to pick up, know where the claim has got to, and do what they said they would do.

And the delay is not a soft cost. When the Financial Ombudsman examined buildings insurance complaints, claim delay was the second biggest cause, and delay complaints were upheld 57% of the time. More often than not, the delay is judged to be the firm's fault.

What a claims-capable agent should actually do

  • Answer, at the hour the claim happens. Fires and floods do not respect office hours.
  • Take the details properly, once. Policy number, what happened, when, where, photographs. The point is that nobody has to give the same information twice.
  • Confirm what happens next, in plain language, so the customer is not left guessing.
  • Chase the outstanding documents without a human having to remember.
  • Give a status update when the customer rings again, which is the single most common reason they do.
  • Hand over the moment judgement is needed. Whether to claim, what the excess means for them, whether cover applies: those need a person.

What it must never do

  • It must not advise. The FCA is clear that advice "requires an element of opinion" and amounts to "a recommendation as to a course of action", while information "involves statements of facts or figures". Telling a customer whether to claim is an opinion. Telling them what their policy says is a fact.
  • It must not decide the claim. Cover decisions are not an agent's to make.
  • It must not guess. If the answer is not in your knowledge base, it says so and passes over.
  • It must not trap the customer. A distressed claimant who asks for a person gets a person.

The Consumer Duty angle, since it applies here too

The FCA's guidance on the Duty specifically warns that "the means of making a claim should be easy to find and the firm should not impose unreasonably restrictive, rigid or arbitrary administrative requirements on customers that create barriers to them making a claim".

An AI agent that makes it easier to notify a claim, at any hour, is squarely on the right side of that. An AI agent that becomes another barrier between a claimant and a human is squarely on the wrong side, and no amount of technology will save the firm that deploys it that way.

Who is accountable

You are. The FCA has confirmed it is not writing a separate AI rulebook: in April 2026 its chief data, information and intelligence officer said plainly, "on AI policy: no new rules". The agent is judged against the rules you already live under, and outsourcing the conversation does not outsource the responsibility.

So insist on the things that let you supervise it: an approved knowledge base you own, a defined escalation route, and a full, readable transcript of every conversation. If the agent said it, your firm said it.

Frequently asked questions

Can an AI agent take a first notification of loss?

It can capture and pass on the claim details inside your firm's permissions and process. Notifying a claim and supporting it with evidence is a regulated activity in its own right, so it happens under your authorisation, not around it.

Can it tell a customer whether to claim?

No. That is an opinion, and opinions are advice. It gives facts and hands the customer to a person.

Can it decide whether cover applies?

No. Cover decisions belong to your firm and its people.

Does an AI agent create new regulatory obligations?

No. The FCA has said there will be no new AI rules. The existing ones apply, and your firm stays accountable for what the agent says.

Sources

See it answer for your own business

Tell us how you run, and hear it answer live, built around your own information.

See pricing