Aipedia: Insurance brokers

AI fact-finds and the demands and needs test

Aipedia, by the Apex broking team · Last reviewed 2026-10-27

A broker can use an AI fact-find, but ICOBS 5.2.2R still requires the firm to specify the customer's demands and needs from what the customer tells it. If the AI asks the wrong questions, the gap in cover is the broker's problem.

Part of: AI in insurance brokers

In short

AI fact-finds adapt their questions to the client's answers, summarise discussions and suggest products. Under ICOBS 5.2.2R(1) the firm must specify the customer's demands and needs from information obtained from the customer, and under ICOBS 5.2.2BR any contract it proposes must be consistent with them. A question set that never asks about a key exposure can leave the client underinsured, and the resulting claim is usually framed as broker negligence. Broker E&O cover will commonly respond, subject to its terms, but prevention is cheaper.

How brokers use AI in the fact-find

Brokers use AI at the start of the client journey in several ways. Dynamic question sets change the next question based on the last answer, so a trade client is asked about subcontractors but a shop is not. Large language models summarise a discovery call into a demands and needs statement. Some platforms read a client's website or accounts and suggest the covers the business probably needs. For retail customers, online journeys may complete the whole fact-find without a member of staff.

These tools save time on simple risks and can prompt a broker to raise covers they might forget. The weakness is that a model only asks about what it was built or trained to expect, and a summary is only as good as the conversation it summarises.

The duties that apply

ICOBS 5.2.2R(1): before the contract is concluded, the firm must specify, on the basis of information obtained from the customer, the customer's demands and needs.
ICOBS 5.2.2R(2) and (3): the level of detail must reflect the complexity of the contract and the type of customer, and the statement must be communicated before conclusion of the contract.
ICOBS 5.2.2BR: any contract proposed must be consistent with the customer's demands and needs.
Consumer Duty: for retail customers, PRIN 2A.2.8R requires firms to avoid causing foreseeable harm, which includes selling cover that does not fit.
SYSC 8.1.6R: where the fact-find tool is supplied by a third party, the firm remains fully responsible.

The demands and needs statement is a regulatory record as well as a sales document, so an AI summary that misrecords what the client said can undermine the broker's defence later.

What can go wrong

Illustrative: a small manufacturer completes an AI-led commercial fact-find. The question set was configured for office-based businesses and never asks whether the client exports products. The broker arranges a combined policy with territorial limits that exclude the client's main overseas market. A product liability claim arrives from abroad, the insurer declines, and the client alleges the broker failed to identify an obvious need.

Other common failure modes include an AI summary that drops a client's stated concern, a model that steers every client towards the same product, and a retail journey that cannot record an unusual circumstance because there is no field for it.

How PI usually responds

Claims about failing to identify needs are a familiar type of broker E&O claim, and the broker E&O claims examples page shows how they arise. Broker PI must meet the MIPRU 3.2.4R minimum terms, including cover for the conduct of the firm and its employees. An error made through an AI fact-find will usually be treated as an error in the conduct of the business, subject to the policy wording and any AI exclusion or condition. The client's loss is often measured by what a suitable policy would have paid, which can be substantial.

Practical controls

Map each AI question set against the exposures a competent broker would check for that trade or customer type.
Require a person to review AI-generated demands and needs statements for commercial risks before quoting.
Give clients a free-text route to tell you anything the questions missed.
Keep the question set version and the client's raw answers on file, not just the summary.
Test outcomes periodically: are clients with similar profiles getting similar recommendations?
Record which supplier provides the tool and how it is monitored.

How Apex helps

Apex arranges PI for insurance brokers. A broker can review your E&O cover and check how the wording treats advice and fact-finds produced with AI tools; see the insurance brokers PI guide.

Sources

Frequently asked

Can an AI tool complete the demands and needs statement on its own?

ICOBS does not prohibit automation, but the firm must still specify demands and needs from information obtained from the customer and ensure any proposed contract is consistent with them. For complex commercial risks, a human review before quoting is a sensible control.

Who is liable if a third-party fact-find tool misses a question?

As far as the FCA is concerned, the broker. SYSC 8.1.6R keeps the firm responsible for outsourced functions. The broker may have contractual rights against the supplier, but the client's claim will usually be made against the broker.

Does the demands and needs statement need to show AI was used?

ICOBS does not specifically require that. It is still good practice to record how the fact-find was carried out, including the tool and version, because that evidence helps the firm explain its advice if a complaint or claim follows.

Check your PI covers how you use AI

Tell us about your firm and the work you do. A broker will review how your current PI treats AI-assisted work, or arrange new cover. No fee for a quote. Or call 0117 325 0027.

Start a PI quote Request a callback

Apex Insurance Brokers Limited is authorised and regulated by the Financial Conduct Authority. Registered in England and Wales, company number 07014570. This page is general information about how AI is used in the profession and how insurance may respond. It is not legal, regulatory or insurance advice on your circumstances, and it does not guarantee that cover will be available or on what terms.