How to choose an AI consultancy in Yorkshire

There are a lot more AI consultancies in Yorkshire than there were two years ago. Some of them are very good. Some of them were web agencies in 2024, crypto agencies in 2022 and will be something else by 2028. From the outside, the websites look much the same.

We are one of the firms you’d be choosing between, so read this with that in mind. But we’d rather you hired a good competitor than a bad anyone, because the bad ones are what give the whole trade a name. Here’s how we’d pick.

First, be clear what you’re buying

“AI consultancy” covers three quite different jobs, and most firms are only honest about doing one of them well.

Advice. Someone looks at how your business runs, tells you where AI would help, where it wouldn’t, and what it would cost. The output is a plan.

Building. Someone takes a specific job (quoting, onboarding, chasing invoices, triaging enquiries) and builds the tool or automation that does it. The output is working software.

Training. Someone gets your staff using the tools that already exist, properly and safely. The output is people who know what they’re doing.

The ONS found in July 2026 that lack of expertise is the most-cited barrier to AI adoption among UK firms that report one, and that only 11% of businesses have trained more than half their workforce (Artificial intelligence in UK businesses, ONS, 20 July 2026). So the third job is more valuable than it sounds, and the first is often a disguised sales pitch for the second. Decide which you need before you ring anyone.

Eight questions worth asking

1. “Show me something you built that’s still running.” Not a demo. Not a slide. A thing a real business uses every day, with the client’s permission to talk about it. Ask how long it’s been live and what’s gone wrong with it since. A consultancy that’s never had anything go wrong has never shipped anything. Our own are on the work page, warts included.

2. “What would you tell me not to do?” The single most reliable test. A good consultant will, within ten minutes, identify something you were hoping to automate and tell you it isn’t worth it. If everything you mention is “a great use case”, you’re talking to a salesman.

3. “Who actually does the work?” Ask whether the people in the pitch are the people who’ll build it. Ask if any of it is subcontracted, and to whom. Ask whether they’re in Yorkshire or whether “Yorkshire” is a landing page.

4. “Where does my data go?” They should be able to answer, specifically, which AI providers your information will pass through, under what terms, whether it’s used for training, and where it’s stored. If the answer is a shrug or “it’s all encrypted”, that’s a no. The ICO’s guidance on AI and data protection is the baseline, and a consultancy should know it well enough to tell you what it means for your job.

5. “What happens when it’s wrong?” Every AI system is wrong some of the time. The question is whether the design assumes that. Ask what the tool does with a case it can’t handle, who gets told, and how you’d notice if the error rate crept up. We wrote about why this matters in what AI agents can actually do.

6. “What do I own at the end?” The code, the prompts, the automation scenarios, the accounts they sit in. Insist that it’s built in your accounts, not theirs, so if you fall out you keep the thing. A surprising number of businesses discover they’ve been renting their own operations.

7. “What will it cost to run, not just to build?” Model usage, tool subscriptions, hosting, and the maintenance that every piece of software needs when a supplier changes an API. Get the monthly figure in writing. We’ve laid out our own ballpark in what AI automation actually costs.

8. “Can we start small?” The right answer is yes. One process, a few weeks, a fixed price, and a clear measure of whether it worked. Anyone insisting on a six-figure “transformation programme” before they’ve automated a single invoice is selling the programme, not the result.

The red flags, in one list

• A proposal that arrives before anyone has asked how your business works.
• The word “transform” used more than once.
• No named person who’ll be accountable after launch.
• A reluctance to put you in touch with a past client.
• Pricing that is only available as a percentage of “projected savings”.
• An insistence that you need a chatbot. Most businesses don’t, or not first.
• Any mention of “AI-powered” without being able to say what the AI does that a normal bit of software couldn’t.

Why local still matters

It’s all remote now, and you can hire a consultancy in Lisbon as easily as Leeds. Three reasons to look closer to home anyway.

The first is that the useful part of the job happens in your building. Watching how an order actually gets processed, who re-types what, where the paper lives. That’s a morning in Huddersfield, not a Teams call.

The second is accountability. A firm whose reputation lives in the same business community as yours has more to lose from a bad job.

The third is simply that Yorkshire businesses tend to have Yorkshire problems: manufacturing, logistics, care, construction, professional services, and a long tail of family firms that have run the same way for forty years and now need to not. A consultancy that mostly serves London fintechs will not recognise your business when it sees it.

What we’d say about ourselves

We’re a small Yorkshire firm that builds things: automations, AI agents and bespoke software, for businesses across Leeds, Sheffield, Bradford, Wakefield, York and the rest of the county. We’ll do the advice bit, but we’d rather be judged on what’s running. We’ll tell you when summat isn’t worth doing. And everything we build sits in your accounts, not ours.

If you’d like to ask us the eight questions, get in touch. We quite enjoy answering them.