A good AI consultant finds where AI creates real value in your business and implements it — training, automation, or custom builds — then proves it with measurable results. The ones to avoid sell demos and jargon; the ones to hire ship working systems and let you call their references first.

The skepticism around AI consulting is loud, and honestly, it’s earned. “It feels like all the crypto hype but to an even more extreme degree,” one commenter wrote. Another was blunter about the vendors: “Remember, the AI companies have something that is able to do interesting product demos — but that is it.”

And the sharpest critique cut at the inflated promises directly: “Claims feel a little inflated — ‘10+ hours saved’ — and that hurts credibility everywhere. How do you know it’s 10 hours when you don’t even convince me you know what I spend 10 hours a week on?”

That last one is the why behind this whole post: the fastest way to spot a real AI consultant is that they understand your work before they promise to fix it. Here’s how to tell substance from spin — and what to ask before you hand anyone a check.

What does an AI consultant actually do?

Cut through the title. A real AI consultant does four things, in roughly this order: figures out where AI actually creates value in your business, implements it (training, automation, integration, or a custom build), proves it saved time or made money, and leaves your team able to keep going.

The key word is implements. The difference between a useful consultant and an expensive one is whether they ship working systems or hand you a strategy and walk away.

How do you separate substance from hype?

A few reliable tells:

  • They ask before they pitch. Substance starts with your workflows, your tools, your week. Hype starts with a demo reel.
  • They’re specific about limits. A trustworthy consultant will tell you what AI can’t do for you yet. Vendors selling “extreme crypto-grade hype” never mention a downside.
  • Their claims are grounded. “We’ll save you 10 hours” means nothing unless they can point to which 10 hours, in your actual operation. Grounded beats grandiose.
  • They show working code, not just slides. Demos are easy. A running automation in your environment is not.

The questions to ask before you sign

Bring these to the first call. The answers separate the field fast:

  1. “What’s the first workflow you’d automate, and why that one?” — tests whether they understand your business or just their product.
  2. “What will I have actually working at the end of this engagement?” — surfaces deck-makers immediately.
  3. “How will we measure whether it worked?” — a real consultant wants a number; a hype merchant changes the subject.
  4. “Can I talk to three past clients before we start?” — the single best filter. The confident ones say yes, now, not after you sign.
  5. “What won’t AI fix here?” — the most revealing question you can ask. An honest “here’s where it won’t help” earns more trust than any promise.

What a good engagement looks like

A working result, not a binder. Even a first, fixed-scope engagement should leave you with at least one automation or tool live and in daily use, your team trained to extend it, and a clear measure of what it saves. Favor fixed scope tied to an outcome over open-ended hourly billing — so you see value before you commit to more.

A caveat worth stating plainly: no consultant, including us, can promise AI will transform your business overnight. Anyone who does is the hype the skeptics are warning you about. What a good one can promise is a concrete first win, measured honestly, with references who’ll vouch for it.

Ask for the working system and the phone numbers. The right consultant hands you both before you’ve signed anything.