Most brokers didn't get into mortgages to spend their evenings writing file notes. Yet a lot of a working week goes on the paperwork around advice: chasing payslips, rewriting suitability letters, keeping the CRM tidy and reminding clients that their fixed rate is ending. This guide covers where AI helps a UK mortgage broker, where the regulator draws the line and how to start without putting your permissions at risk.
Why are UK mortgage brokers looking at AI now?
The intermediary channel is already dominant. In its December 2024 report The new "normal": prospects for 2025 and 2026, IMLA estimated that around 87% of mortgage business went through intermediaries in 2024. As more cases move through brokers, more fact finds, document chases and product transfer reviews land with the same small teams.
The wider industry has already started using AI. A joint survey by the Bank of England and the FCA, published in November 2024, found that 75% of responding financial services firms were already using AI, with a further 10% planning to do so within three years.
What does the FCA actually say about AI?
The clearest recent statement is the Mills Review, which the FCA published on 6 July 2026. It looks at how AI could reshape retail financial services and draws on a survey of more than 5,000 consumers. The regulator's chair said its principles-based approach, built on the Consumer Duty and the Senior Managers regime, was the right one. The review makes no mortgage-specific recommendations.
In practice, the rules you already follow apply to AI output too. If a tool drafts a suitability letter that misstates a client's income, it's still a suitability failure. The tool has no permissions and no accountability. You have both.
Which broker tasks suit AI best?
Here are some concrete examples from a typical case file:
- File notes. Dictate for two minutes after a call, or feed in the call transcript, and get a structured note back with objectives, circumstances, preferences and next steps. The adviser corrects it and saves it.
- Suitability letter drafts. Give the tool your firm's approved template and the adviser's reasoning, and ask for a first draft in plain English. The adviser edits the recommendation section by hand.
- Document chasers. Draft polite, specific emails listing exactly what's missing, for example the last three months of statements for the joint account, rather than a vague "please send documents".
- Product transfer reminders. Draft a sequence of messages timed to a client's rate expiry. The same pattern that works for automated appointment reminders works here.
- Paperwork summaries. Summarise long bank statements or self-employed accounts so the adviser knows where to look. The adviser still reads the source documents.
Other professional firms have found the same split. The approach set out for accountants and solicitors using AI maps closely onto brokering: let AI take the first draft and keep professional judgement with a person.
Where should brokers not use AI?
General-purpose chatbots don't have live access to lender criteria, sourcing systems or product changes. Ask one whether a lender accepts a particular income type and you may get a fluent, plausible and outdated answer. Your sourcing system and the lender's own criteria remain the source of truth.
Vulnerability needs particular care. Under the Consumer Duty, spotting and responding to vulnerable customers is a human judgement that depends on context, tone and follow-up questions. An AI summary can drop the one remark that mattered, such as a mention of a recent bereavement or a health condition.
How do brokers protect client data when using AI?
Mortgage files hold some of the most sensitive data a small business ever touches: income, debts, credit history, sometimes health details. Pasting that into a free consumer chatbot account is where most avoidable risk comes from. Before approving any tool, answer four questions in writing: where is data stored, is it used for training, can you sign a processing agreement, and who in the firm can use it.
For a fuller walk-through, work through this AI data protection checklist for UK businesses before anyone uploads a real client file.
Do brokers need to understand how their AI tools work?
The same Bank of England and FCA survey found that 46% of firms reported only a partial understanding of their AI technologies, compared with 34% reporting a complete understanding. It also found that a third of AI use cases were third-party implementations, which is the position most brokers are in when they buy a tool rather than build one.
You don't need to know how a model works internally. You do need a one-page policy that covers which tools are approved, which tasks they may be used for, who reviews the output and how that review is recorded on file.
How should a small brokerage start with AI?
A practical rollout looks like this:
- Choose the task. File notes are the usual winner because every case needs them and every adviser dislikes writing them.
- Approve the tool. Use a business plan, check the data terms and log the decision.
- Build the template. Write the exact headings your compliance checks expect, so the output lands in the right shape.
- Run a pilot. One adviser, four weeks, a simple log of minutes saved and errors caught.
- Review with compliance. If you're an appointed representative, speak to your network before going further, because many networks set their own AI rules.
- Scale carefully. Add the next task only once the first one is stable.
What about client-facing chatbots on a broker website?
The Mills Review found that a fifth of consumers, equivalent to around 11 million UK adults, would be likely to use AI that acts autonomously within pre-set goals, so clients will increasingly arrive having already asked a chatbot about their mortgage. A good website assistant meets them there, answers questions such as "what documents will I need", and books them in with a person. Estate agents face a similar balance, and the guide to AI for estate agents covers the handover to a human in more detail.
Which AI tool should a mortgage broker use?
Specialist mortgage platforms are adding AI features to CRMs and sourcing systems, and those are worth using where they fit your existing workflow. For everything around those systems, including letters, emails, notes and summaries, a general assistant set up with your firm's templates and instructions covers most of the work. The Claude hub explains how that setup works for small UK firms.
AI for UK Mortgage Brokers — FAQ
Is it legal for a UK mortgage broker to use AI?
Yes. No FCA rule bans a mortgage broker from using AI. In its Mills Review, published on 6 July 2026, the FCA did not propose AI-specific regulation and said it would keep relying on the Consumer Duty and the Senior Managers regime. That makes the firm answerable for any output AI helps produce. If an AI-drafted suitability letter gets a client's circumstances wrong, the adviser who signed it and the firm that sent it own the mistake. The safe approach is to treat AI as a drafting and admin tool, keep a qualified adviser accountable for every recommendation, and write down how the tool is used, checked and supervised.
Can AI give mortgage advice to my clients?
Not on its own, and not under your firm's name without a qualified adviser checking it. A regulated recommendation has to reflect the client's full circumstances, affordability and needs, and the Consumer Duty expects good outcomes you can evidence. General AI tools can get lender criteria wrong, miss product changes and state guesses with confidence. Use AI to prepare: summarise the fact find, draft questions, turn call notes into file notes and write the first version of letters. The adviser makes the recommendation and signs it off. Where AI output shapes advice, keep a record showing a human reviewed it.
What are the best first tasks to hand to AI in a brokerage?
Start with repetitive, low-risk writing where a human already checks the result. Good candidates are file notes from call recordings or dictation, first drafts of suitability letters built from your own templates, chasing emails for missing documents, product transfer reminders and summaries of long bank statements or payslips for the adviser to verify. These tasks eat hours, follow a pattern and are easy to check line by line. Leave lender selection, affordability judgements and anything to do with vulnerable customers under full human control until your review process is proven.
How widely is AI already used in UK financial services?
A Bank of England and FCA survey published in November 2024 found 75% of responding financial services firms were already using AI, with a further 10% planning to within three years. The same survey found 46% of firms said they had only a partial understanding of the AI they use. For a broker, that second figure matters more. Adoption is common, but many firms cannot fully explain how their tools reach an output. Before rolling anything out, make sure you know where client data goes, what the tool can and cannot do, and who checks its work.
Do I need client consent to use AI on their data?
UK GDPR still applies, so you need a lawful basis for processing, a privacy notice that explains how data is used and suppliers that meet your data protection duties. Check whether the AI provider stores or trains on your inputs, where data is held and whether you can sign a data processing agreement. Business plans from the major providers usually offer stronger terms than free consumer accounts. Update your privacy notice to mention AI-assisted processing, keep special category data such as health details out of tools that are not approved for it, and record your assessment in case a client or the ICO asks.
