Find the one thing that is actually costing you money. Then fix it.
Most AI advice starts with the software. I start with your week — where the hours go, where work gets stuck, and which single bottleneck is worth four weeks of effort. Then we point the AI you are already paying for at that one thing.
Around one in five UK small businesses use these tools regularly. Roughly six in a hundred have them properly built into how the business runs.
The firms that started two years ago mostly spent the first eighteen months on things that went nowhere. Starting now, with someone who already knows which parts to skip, is a better position than they had.
The window is open. It is not going to stay open once your competitors work this out.
For twenty years, the good tools were not sold to you.
Serious business software came with a six-figure licence, a mandatory implementation partner, an eighteen-month rollout and a consultant who spoke in acronyms. If you run a business with nine people and a van, you were never the customer. Your options were a spreadsheet, a cheap tool that half fitted, or doing it by hand.
That is the part that changed. Microsoft and Google have now folded AI into the price of the subscription you already pay — which means the expensive part is sitting in your business right now, switched on, doing nothing.
I spent fifteen years on the inside of the old world. I know which parts of it you can now skip.
Where, what it is worth, and what to leave alone
Where the problem actually is
Usually not where it feels like it is. Owners tell me a job takes three days. It rarely does — most go out fast and a minority drag badly, and it is the minority costing the money. Averages hide that. Looking properly at how work moves through your week is what finds it.
What fixing it is worth
Real numbers against your own figures: hours recovered, work won instead of lost, what that is worth a year, and how long before it pays for itself. Including which parts of that estimate are soft and need checking. If the number does not work, I tell you it does not work.
What to leave well alone
Some of what these tools could do to your business, they should not. Anything carrying regulatory exposure, anything where being wrong costs you a customer relationship, anything that takes a year to find out whether it worked. You get that list too, with reasons.
One page. Not a forty-slide deck.
Everything I find fits on a single side of paper, in the language you use, with the numbers attached. You can read it in four minutes and hand it to whoever else needs to see it.
It is yours whether or not you take it any further. Plenty of businesses take the page and do the work themselves. That is a perfectly good outcome.
QUOTING TURNAROUND — WHAT THE DATA SHOWS
Experimenting is cheap now. That is the whole point.
No rollout, no migration, no eighteen-month programme. One small change, measured, kept only if it earns its place.
Walk me through last Tuesday
At your premises. You talk about how the week actually runs. I ask awkward questions and we agree what is worth a closer look.
The diagnosis
I work through what your own records already show, mostly sitting with you rather than taking anything away. Ends with the page.
Worth doing, or not
Some problems are not tool problems. If yours is about pricing, capacity or a system that needs replacing, I say so and we stop there.
One change, handed over
Two hours with the people who do the work, then build it on your real cases. You get written instructions, a named owner inside your team, and a before-and-after number. Then I leave.
Plain, straight to camera
Based in [YOUR AREA]. Working with businesses across West London and the Thames Valley — close enough to sit in your office on a Tuesday afternoon.
I am not an AI consultant. I work out how businesses get stuck.
I am [YOUR NAME]. Fifteen years as a data product owner at [YOUR EMPLOYER], responsible for how information is structured, connected and made usable by the systems — and now the AI tools — that sit on top of it. Incomplete records arriving at a boundary and blocking everything downstream is a problem I have solved a hundred times. It just has pallets in it this time.
I have sat through the vendor pitches. I know which promises survive contact with real work and which quietly disappear after the invoice. I have watched companies with enormous budgets fail at this for reasons that had nothing to do with money.
I do not resell software, I take no commission from anyone whose product I suggest, and I am not trying to become a line on your monthly bills.
Prices, before you ask for them
Published because you should not have to sit through a discovery call to find out whether you can afford it.
The first hour
A conversation at your premises. If I cannot see anything worth looking at, I say so and we both save the time.
The diagnosis
The bottleneck found, evidenced and costed, on one page. Yours to keep and act on yourself. Most businesses should start and stop here.
One change, built
Four weeks, one process, built on your real cases and handed over with instructions. Fixed price agreed up front. No retainer afterwards.
The expensive part — the AI itself — is almost certainly already inside the subscription you pay for. What is sometimes missing is a small amount of extra licensing to connect it to the way you actually work. That is usually tens of pounds a month, not a project, and I will tell you the number before you commit to anything, not in week three.
The things owners actually say to me
Am I too late?
No. Around a fifth of UK small businesses use these tools regularly and only a handful have them properly embedded. Most of the early movers wasted eighteen months. Starting now, with someone who knows which parts to skip, is a better position than they had.
I am not handing my customer data to a stranger.
Good. You should not. Most of what I need is timings, not content — when work arrived and when it went out, with no names, prices or attachments. Where I do need to read the detail, I do it sitting at your desk and nothing leaves the building. Signed confidentiality and data agreements before anything starts.
My records are a mess.
Everyone says this and everyone is right. Making sense of messy information is the actual job I have done for fifteen years. We work with what you have rather than waiting for a tidy-up that never comes.
What if there is nothing worth doing?
Then I write that down and we stop. Some problems are about price, capacity, or a system that needs replacing, and no tool touches those. You still get the page, which tells you where the money is actually going. That is worth having on its own.
Do I have to replace my staff?
Almost nobody does. The large majority of small firms using these tools report no change to headcount. What changes is how much of the day goes on typing things twice and chasing information that should have arrived complete.
I already have an IT company.
Good, keep them. They look after your machines, your email and your security, and they do it well. This is a different question — what the business should do differently now this capability exists. I am happy to work alongside them and often do.
[WHAT THEY THOUGHT THE PROBLEM WAS, WHAT IT TURNED OUT TO BE, AND WHAT CHANGED][NAME], [ROLE], [BUSINESS AND TOWN]
[IDEALLY ONE FROM SOMEONE YOU TOLD NOT TO PROCEED — THE MOST PERSUASIVE QUOTE YOU WILL EVER GET][NAME], [ROLE], [BUSINESS AND TOWN]
An hour, and you leave knowing where it hurts.
Tell me roughly what the business does and which part of the week annoys you most. If there is nothing here worth your money, you will hear that in the first hour rather than the fourth.