AI and the Broker Business: Faster, Smarter, Still Human-Led

Richard Howes

26 August 2026

“You don’t need Chips with Everything”

In a recent Sunday Times piece, "Will financial advice models survive the great wealth transfer?", the historian and generational change expert Eliza Filby set out the scale of the challenge facing financial advice as trillions of pounds pass from baby boomers to their children and grandchildren. Her argument is not that advice is becoming obsolete. It is that the model has to adapt, because the next generation of clients wants a different relationship: less deference, more transparency, and trust that is earned rather than assumed.

That same tension sits underneath the conversation about artificial intelligence in mortgage and protection broking. AI is not simply a clever tool or a matter of better prompts. It is becoming something people talk to, consult and increasingly rely on, in the same way a new generation of clients is starting to expect more from the advisers they work with. For broker firms, that shift is useful and risky at the same time. Used well, AI can strip out admin, sharpen client communication and free up time for the conversations that matter. Used carelessly, it risks quietly replacing the judgement clients are actually paying for. The real question is not whether AI is useful. It is where it is useful, where it is risky, and what must stay firmly human-led.

Frame 1: AI as sat nav

AI works best as a sat nav for a broker firm: it can make the journey faster and easier, but it should never mean the adviser loses the ability to navigate. Good uses include drafting client emails, preparing meeting notes, summarising lender criteria, building case packaging checklists, helping advisers prepare for client conversations, turning long documents into quick summaries, and producing first drafts of marketing or client education content.

Just as a sat nav can dull a driver's natural sense of direction over time, AI used lazily could dull professional judgement. The question worth asking in your own firm is where AI could genuinely remove friction, and where leaning on it would weaken the judgement clients are paying you for.

Frame 2: The broker value test

The useful question is not "can AI do this?" It is "should AI assist this, produce this, or decide this?" A sensible operating model treats AI as an assistant for repeatable, text-heavy, admin-heavy work. It can produce first drafts of communications, summaries and checklists. It should never own judgement, suitability, client understanding or accountability.

In practice, that means AI can draft a client email, but the broker checks the tone, the facts and the suitability. It can summarise lender criteria, but the adviser verifies it against current lender policy. Put simply, it helps you not starting with a blank piece of paper when starting a job. It can help prepare a case note, but the broker confirms the facts and the client's intent. It can support product research, while the adviser remains accountable for the advice given. Worth asking: which parts of your business would you be comfortable letting AI assist, and which parts must remain firmly adviser-led?

Frame 3: Trust is the asset

For broker firms, trust is the commercial asset, and it is exactly what Filby argues is under pressure as wealth and expectations shift between generations. AI can make a firm look more responsive, polished and efficient. But if it produces something that sounds confident and turns out to be wrong, the damage to trust can be significant and quick.

The risks worth watching for are inaccurate lender criteria, overconfident explanations, generic suitability wording, incorrect assumptions about affordability, a tone that misreads the client, advice-style language that has not been properly checked, and a gradual loss of individual adviser accountability. If AI helps a firm sound more confident, the firm needs a matching check to make sure it is actually more accurate, not just more articulate.

Frame 4: Treat AI as a junior assistant, not a senior adviser

The simplest working rule: AI creates the first draft, humans make the decision, and the firm owns the outcome. That framing keeps AI useful without letting it get out of hand.

Good tasks for AI include first-draft client emails, pre-meeting preparation, follow-up notes, marketing ideas, criteria summaries, internal knowledge summaries, call preparation, and training support for newer advisers. Tasks that must stay human include suitability judgement, final client advice, assessment of customer vulnerability, interpretation of complex circumstances, recommendations, escalation decisions, and handling complaints or sensitive customer issues. If AI joined your firm tomorrow as a very fast but inexperienced assistant, it is worth deciding in advance what you would hand it and what you would never let it do unsupervised.

Where the real value sits

Some questions you could ask yourself and of your business which might help or determine your AI journey:

• “Where could AI genuinely remove friction in your business, and where could relying on it weaken the professional judgement your clients pay you for?
• Which parts of your business would you be comfortable letting AI assist, and which parts must remain firmly adviser-led?
• If AI helps us sound more confident, what checks do we need to make sure we are more accurate?
• Where could AI create genuine business value in your firm, and where might it simply create more polished-looking work without improving outcomes?
• If AI joined your firm tomorrow as a very fast but inexperienced assistant, what would you give it — and what would you never let it do unsupervised?

Not every use of AI creates value. Some of it just produces more polished-looking work without changing the outcome for the client or the business. The firms that get the most out of AI will be the ones that ask, honestly, where it creates genuine business value in their firm, and where it simply makes things look busier.

A good way to start is small. Pick one AI use case you would be comfortable piloting in your firm, whether that is drafting client follow-ups, summarising lender criteria, or preparing meeting notes, and put a clear human check around it before it touches a client. That is the model Filby's argument points to as well: the advice relationship survives generational and technological change not by resisting it, but by staying unmistakably human at the point that matters most.

I am not sure that AI is coming for the broker's job, but perhaps the answer as with many things is not to look at the two extremes of the spectrum perhaps the answer lies in the middle. Used with the right boundaries, it can make good advisers faster and better prepared, freeing up more time for the conversations, judgement and trust that clients actually pay for.