How to choose an AI training provider: a buyer's checklist

Why every provider sounds the same

Start shopping for AI training and you'll notice the pitches blur together. Everyone promises to be hands-on, practical, tailored, and outcome-focused. The words are free, so they're everywhere, which means they tell you nothing. The real differences sit one layer down, in how the work is actually done.

The good news: a handful of pointed questions cut straight through the sameness. If a provider answers them crisply and specifically, they've done this before. If they retreat into adjectives, you have your answer.

The four questions that actually separate providers

Ask these on the first call, and listen for specifics rather than reassurance:

• Who actually delivers it? You want practitioners who build with AI daily, not a sales lead who reads someone else's slides. Ask to meet the actual trainer and what they've shipped.
• How much is tailored to our work? The session should be built around your real tools, tasks, and data, not a standard deck with your logo. Ask what they'd change about the curriculum for your team specifically.
• What happens after the session? Skills fade without reinforcement. Ask what your team leaves with (prompts, playbooks, a follow-up) and what stops it being forgotten by Friday.
• How will we know it worked? A serious provider baselines where your team starts and checks what stuck. If they can't describe how they'd measure it, they're selling a day out, not a capability.

Notice none of these are about the model or the tools. The hard part of corporate AI training was never the technology. It's getting a specific team to change how it works, and that's what these questions probe.

Red flags worth walking away from

A few signals reliably predict a disappointing engagement:

• A fixed price before they've asked about your team. It means a generic session is coming. Real training is scoped first. See our note on what AI training actually costs.
• All theory, no hands-on. If your team won't be building on their own real work during the session, the skills won't transfer.
• No follow-up offered. One-and-done training is the most common way budgets get wasted.
• Proof that's vague or borrowed. "Trusted by big logos" with no detail (or testimonials about unrelated work) isn't evidence they can train your team on AI.

Any one of these on its own is a conversation; two or more together is usually a reason to keep looking.

How to make the final call

Once you have a shortlist, stop comparing proposals and test the real thing. Run a single paid pilot session with the top one or two providers, with a real team, on real work. You'll learn more from one session (how the trainer handles questions, whether people are still using what they learned a week later) than from any number of capability decks.

Score the pilot on the four questions above plus one more: did the room actually change how it works afterwards? Whichever provider moves that needle is the one to scale with. The decision also depends on whether you'd be better served building the capability internally instead. We weigh that up in in-house vs outsourced AI training. If you want to run that pilot with us, our AI training is built to be scoped to exactly your team, from five people to a full enterprise rollout. For a worked example of what a vertical brief should cover, see AI for accountants. The six tests UK evidence proposes for judging any provider are in AI training for business.