What changes
In most broker firms, AI use today is patchy. One person drafts emails with a chatbot, someone else won’t touch it, and nobody is sure what client data is allowed near it. Building capability means the whole team learns the same basics, works to the same rules and shares what works. The aim is people who understand AI well enough to use it every day and to catch it when it gets something wrong.
This matters more in mortgages and protection than it might seem. An AI tool can misread a payslip, invent a lender criterion or summarise a medical disclosure badly. The adviser who signs off the recommendation is still responsible, so the skill that counts most is checking.
Why it matters now
The next moves depend on this one. You can’t hand over the legwork, build your own client tools or set fair prices for an AI-assisted service if your team doesn’t understand what the technology does well and badly. Buying a tool without that understanding usually ends in a subscription nobody uses, or in one that’s used without proper checks.
Checklist
- Put everyone through the same training. Advisers and administrators both, starting with the fundamentals and practical use, so the team shares a vocabulary.
- Write a one-page AI use policy. Cover which tools are approved, what client data may go into them, and the rule that a person checks every output before it reaches a client or lender.
- Build a shared prompt library. Start with five tasks from your Move 1 list, such as summarising a fact-find, drafting a protection needs explanation or turning bank statements into a spending summary.
- Hold a weekly half-hour session. One person shows something that worked and one thing that went wrong. Keep notes, because the mistakes teach the most.
- Check data handling with your compliance support. Confirm how your approved tools store and use data, and record the decision.
Illustrative example
Illustrative example: a four-person firm (two advisers, two administrators) agrees a policy and a shared library of eight prompts in its first month. In week three, an administrator notices that the AI summary of a self-employed client’s accounts has used the wrong tax year. The team adds a line to the prompt asking the tool to state the period for every figure, and adds a check to the file review. Nothing reached a lender, and the whole team learned where to look.
Common mistakes
- Training the advisers and leaving out the administrators, who do most of the work AI will help with first.
- Letting people paste client documents into free consumer tools without checking how that data is stored and used.
- Treating AI output as finished. Every figure, criterion and recommendation needs a person to check it against the source before it goes anywhere.
Tools change month to month, so keep the weekly session running once the formal training ends, and review the policy each quarter.