Plenty of AI consultancies will happily take a Welsh business's money over a video call. Some AI work genuinely does run fine remotely, so that is not automatically a problem.
But a fair bit of it, specifically the part that actually changes how your business runs day to day, goes better with somebody who can stand in the room. If you are weighing up an AI consultant in Cardiff or anywhere across South Wales, here is what worth paying for actually looks like.
Why local still matters for this kind of work
The build itself can happen anywhere. A model does not care which side of the Severn it runs on. Understanding your business usually cannot happen anywhere, at least not as quickly or as accurately.
A consultant who can drive to you sees how the work really flows. The workarounds nobody mentions because they have stopped noticing them. The bottleneck that never comes up on a call because it does not feel like a problem until you watch it happen three times in an afternoon. The bit of paper that gets carried from one office to another because two systems do not talk.
You can get at some of that remotely. You get at it faster in person. Speed matters here, because the discovery phase is the part clients are least willing to pay for and the part that most determines whether the result fits.
Newport to Cardiff is half an hour. Swansea, Bridgend, Bristol and Gloucester all sit inside an easy morning. When a job needs somebody on site, that closeness is the difference between a tool that fits your operation and one that very nearly does.
There is a second reason, less technical. Businesses here tend to buy from people they can place. Somebody who knows the difference between Cwmbran and Caerphilly, who understands that a family firm in the valleys runs on different assumptions to a startup in a co-working space, is easier to trust. That is not sentiment. It shortens every conversation.
What good actually looks like
A few things separate practical help from hype. None of them are technical.
One person from first conversation to handover. Not a salesperson who charms you, then an account manager who sits between you and whoever is actually doing the work. You should be able to ask the person who built the thing why they built it that way.
Tools built around your own data. Not a generic template with your logo on it. If a consultancy can deliver the same thing to you as to a firm in a completely different trade, they have not understood either of you.
Somebody who will tell you no. This is the real test. A consultant who says in the free consultation that AI will not help with the thing you asked about has just cost themselves a sale to save you money. That is the person worth hiring.
A clear answer on ownership. When the engagement ends, what do you have. Who can change it. What happens if you fall out with the consultant, or they retire, or they get hit by a bus. If the honest answer is that everything stops working, you have not bought a tool, you have rented a dependency.
If a consultancy cannot do those four things, distance is the least of your problems. A brilliant local consultant who fails on ownership is worse than a competent remote one who does not.
The questions worth asking on the first call
You do not need to understand the technology to run a good first conversation. You need to understand your own business, which you already do. A few questions do most of the work.
Ask what they would look at first, before they know anything about your operation. If they can answer that specifically, they have a method. If they answer with a product name, they have a product.
Ask what the smallest useful version looks like. Good AI work usually starts smaller than clients expect. Anybody proposing to transform everything at once is either inexperienced or selling.
Ask how you will know whether it worked. If nobody can name the number that should move, nobody will be able to tell later whether the money was well spent.
Ask what they will not do. Every honest consultancy has a boundary somewhere.
Where we work
We are based in Newport. We cover South Wales in person, which means Cardiff, Swansea, Bridgend, Caerphilly, Cwmbran, Pontypridd, Merthyr, Monmouth, Chepstow, plus the length of the M4 corridor. Across the border, Bristol and Gloucester are a morning rather than an expedition.
Mid Wales and North Wales, so Aberystwyth, Wrexham, Bangor, everywhere in between, we work remotely, then travel when the job genuinely warrants it. That is a practical judgement made per job rather than a policy.
Distance has never been the deciding factor in whether a piece of work goes well. The work is. A clear problem, an owner who wants it solved, somebody honest doing the solving. That combination succeeds at any range, though it does move faster over a short one.
That local, no-nonsense approach is exactly how we work as an AI consultant in Cardiff and South Wales.
The businesses we tend to know best
Our roots are in the trades that keep the country running rather than in software.
Hospitality and catering, through our sister brand On The Pass Consultancy. Field service and trades, through TorquePlan, which is why the AI for field service side of what we do is written in vans and jobs and callouts rather than in workflow diagrams.
Those are Welsh SMEs in the main. Family firms. Businesses three generations deep, where the person who founded it still has an opinion. They face genuinely different pressures to a venture-funded startup. Tighter margins, longer relationships, more of the knowledge held in people's heads rather than in a system, less appetite for gambling on something unproven.
They also need different answers. A recommendation that assumes a technology budget and an in-house developer is useless here, however good it looks on paper.
The underlying work translates well beyond those two sectors. If your business runs on hard-won knowledge plus too much paperwork, there is a bridge worth building, whatever the trade.
A note on the AI in Wales conversation
There is a good deal of noise at the moment about Wales and artificial intelligence. Growth zones, funding announcements, strategy documents. Some of it will matter to a twenty person firm in Newport. A fair amount of it will not, at least not directly, at least not this year.
The useful filter is simple. If an announcement does not change what you could do next month, it is context rather than news. Worth knowing, not worth acting on.
What does change things next month is usually smaller and much less exciting. The quote that takes forty minutes to produce and could take five. The rota built on somebody's memory of last August. The report the accountant asks for that nobody enjoys assembling. That is where the return is. It has been sitting there the whole time.
What the first month usually looks like
It is worth knowing what you are actually buying, because "AI consultancy" covers an enormous range of things and some of them are not worth much.
The first week is looking. Site visit, a walk through how the work moves, a lot of questions that will feel basic. Who does this. What happens next. Why is it done that way. Most of the value of the whole engagement is decided here, which is why doing it over a video call is a compromise rather than an equivalent.
The second stage is arithmetic and honesty. Which of the jobs we found is worth attacking first, what it costs you now in real hours, what a realistic improvement looks like. This is the point where a good consultant will rule things out. Some of what you hoped for will not be worth doing, at least not yet, at least not first.
Then something small gets built, tested against real work rather than a demo, put in front of the people who will actually use it. They will find problems. That is the point of doing it early.
Finally, training plus handover. Your team needs to be able to run it without ringing anybody. If they cannot, nothing has really been solved, the dependency has just moved.
None of that requires you to understand the technology. It requires somebody who understands your business, then translates in both directions. That is the entire job.
Where to start
If you are curious rather than convinced, that is the right footing.
Book a free consultation. No pitch, no obligation. We will look at where your hours actually go, then tell you honestly where AI would help your business and where it would not. If it makes sense after that, we run a formal efficiency analysis and put real numbers against it.
If you would rather have a look yourself first, our piece on running a bottleneck audit on your own business walks through the same first week we would, without anyone visiting.
If it does not make sense, you will hear that as well. That costs you an hour and saves you a great deal more.