Can AI help mortgage brokers?

Five hours a week is roughly an hour a day. Whether software can hand that back to you depends entirely on where your hours currently go, and most brokers do not know. They have a strong feeling about it, which is not the same thing.

So this is an arithmetic question before it is a technology question. Below is a way of working out whether the five hours exist in your week at all, and which of them are the kind that software can plausibly take.

Find out where the week actually goes

Spend two weeks keeping a rough tally. Not a formal time sheet, which nobody sustains. A note on your phone with a handful of categories and a mark every time you switch.

Useful categories for an adviser: client conversations, writing up notes, chasing documents, chasing lenders, producing suitability documents, internal admin, and answering things that could have been answered by someone else.

Two weeks is enough to see the shape. What usually emerges is that the visible work, the advice conversations, is a smaller share than expected, and that a large block sits in three places: writing things up, chasing things, and repeating information that already exists somewhere in the firm.

Those three are the pool the five hours would have to come from. If your tally says your week is genuinely full of client meetings and complex cases, then no amount of software is going to give you five hours, and the honest answer is that you need another pair of hands rather than another subscription.

Where the minutes are usually recoverable

Write-up after appointments is the biggest single candidate for most advisers. If you conduct six or seven substantive client conversations a week and each takes half an hour to write up properly, that is three hours before you have chased anything. Automatic capture of the conversation does not eliminate that time, but it reasonably turns thirty minutes of composition into ten or fifteen minutes of correction. Call it an hour and a half, if the review discipline holds.

Routine correspondence is the second. Document chasers, appointment confirmations, updates to a solicitor, the same explanation of what a valuation is for. Drafting these from a starting point rather than a blank screen shaves a few minutes each, and the count is high, so it adds up. Somewhere near an hour a week for a busy adviser is a reasonable expectation.

Reading and extracting from documents is the third, though this is the one where the saving is most easily overstated. A summariser can tell you quickly what a long lender criteria page or a client's covering email contains. It cannot be trusted to pull figures for you without checking, and the checking is most of the time.

Internal writing is the fourth and the least glamorous. Meeting notes, process documents, a paragraph for the website, the wording of a template. Small individually, occasionally an hour in a week where several land together.

Where nothing will be saved

Be clear about this, because inflated expectations are why firms abandon tools in month three.

Lender processing time is unaffected. If an underwriter takes eleven days, no assistant software changes that.

Client thinking time is unaffected. People take as long as they take to decide, and hurrying them is not a service improvement.

Anything requiring judgement is unaffected, and attempts to compress it usually create rework. Time spent deciding what to recommend is not overhead. It is the job.

Data entry into lender portals is barely affected in practice. Rekeying is a genuine pain, but it is largely a function of systems that do not talk to each other, and that is an integration problem rather than an AI one.

What the five hours cost you

There is a bill attached, and it is mostly paid in review time and in setup.

Every draft that goes to a client needs a person to read it, and reading carefully takes real minutes. If you save twenty minutes on a suitability-adjacent document and spend fifteen checking it, the net is five. That is still a gain, but it is not the number on the vendor's home page, and a firm that has budgeted for the gross figure will feel cheated.

Setup is the other cost. Getting templates right, agreeing what goes into which system, teaching the team, and fixing the first month of odd outputs is real work. Most firms are net down for the first several weeks.

A realistic answer

For a full-time adviser with a decent caseload, three to five hours a week is a defensible target, and it comes overwhelmingly from write-up and routine correspondence rather than from anything exotic. Firms that report much more than that have usually either measured the gross saving, or fixed a broken process at the same time and credited the software with both.

If you want to test it honestly, do the two-week tally first, pick the single largest recoverable block, change only that, and then tally again after a month. One variable at a time is slower and it is the only way you will ever know what actually worked.

The question underneath

There is a second question worth asking once you have the hours back: what are they for?

If five recovered hours go into taking on more cases, the firm grows. If they go into better file quality, fewer errors and calling clients back the same day, the firm gets stickier. If they go into finishing at a sensible time on a Friday, that is a legitimate answer too.

What tends not to work is recovering the time and never deciding. The hours quietly refill with whatever was already waiting, and nobody can explain what changed.