Questions to Ask an AI Consultant Before You Hire One

A friend forwarded me a proposal last week before she signed it.

Eight thousand dollars. Twelve pages. The word "transformation" showed up four times. Nowhere in it did anyone say what she'd actually have in her hands ninety days later.

She asked me one question: what should I ask before I sign this?

That's really the question underneath every other question people bring me. Not "how much does this cost" or "do I even need this." Those matter, and we've written about both. But the question that actually protects you is simpler: what are the questions to ask an AI consultant before you hire one, so you can tell a real engagement from a very professional-looking guess?

Before you hire an AI consultant, ask about five things: what you'll actually have when the engagement ends, who on their team is doing the real work, how you'll measure whether any of it worked, what happens to your systems after they leave, and where AI has failed for them before. A consultant who answers all five plainly, in plain language, without flinching—that's a good sign. A consultant who gets vague, or moves straight to the contract—that's your answer too.

That's the short version. Here's the long version, question by question, including what a good answer actually sounds like.

And if you'd rather skip the vetting and just talk to someone who'll answer these on a call, that's what our AI consulting work is built around. But bring the list either way. That's the whole point of writing it down.

1. What Exactly Will I Have When We're Done?

Ask this first. Everything else is easier to evaluate once you know the answer.

Why it matters: Vague deliverables are red flag number one. "We'll transform your operations" is not a deliverable. Neither is "AI strategy roadmap" without a noun attached to it. If a consultant can't name the actual thing you'll have—a working workflow, a documented process, a trained team—there's no way to know later whether you got what you paid for.

What a good answer sounds like: Specific and boring, in the best way. "You'll have three automated workflows running in tools you already own, documentation your team can follow without me, and a written readiness assessment showing what to build next." We wrote more on this in the seven things a good AI consultant actually does—deliverables are the first thing on that list too.

2. Who Actually Does the Work—You or a Junior?

This one gets skipped constantly, and it shouldn't.

Why it matters: The senior person pitches you. Sometimes a junior person you've never met does the actual build. That's not always a problem—but you deserve to know it going in, not discover it in week three.

What a good answer sounds like: A name. "I do the strategy and the readiness assessment myself. My associate handles the build, and I review everything before it ships to you." If the answer is a shrug, or "our team," push on it until you get an actual person.

3. How Do You Decide Which Workflows to Automate First?

Ask this before they show you a single tool.

Why it matters: Without a real method for prioritizing, you get whatever tool the consultant happens to be excited about that month—not what actually saves you the most time. Enthusiasm isn't a strategy.

What a good answer sounds like: They describe a process, not a product. Where is time actually bleeding out. What's repetitive, high-volume, and low on judgment calls. What's already broken versus what's just annoying. If you want the plain-language version of what that assessment usually surfaces, we cover the common patterns here.

4. What Tools Do You Recommend—And Do You Take Referral Fees?

Ask both halves of this question. Not just which tool. Whether they're paid to recommend it.

Why it matters: Some consultants steer you toward whatever software pays them an affiliate cut, dressed up as a recommendation. Others just default to whatever's trending on LinkedIn that week. Neither one is thinking about your actual stack.

What a good answer sounds like: Direct disclosure, either way. "I don't take referral fees, so my recommendation isn't tied to anything but fit." Or, if they do: "I get a small affiliate credit from this one, here's why I still think it's right for you." What you're listening for is whether they'll say it out loud without you asking twice.

5. How Will We Measure Whether This Worked?

If nobody can answer this before the work starts, nobody's going to be able to answer it after.

Why it matters: "Efficiency gains" is not a measurement. Without a real number attached before you start, you'll have no way to know—three months in—whether the retainer paid for itself or just kept the invoices coming.

What a good answer sounds like: A number, and a plan for what happens to the time you get back. "You're spending six hours a week on this report. We'll cut that to one, and I want to know what you do with the other five—more client work, more rest, whatever actually matters to you." Time saved that isn't reinvested somewhere on purpose tends to just get eaten by something else.

6. What Happens After You Leave—Can My Team Run This Without You?

This is the question that separates a consultant from a dependency.

Why it matters: If the whole system falls apart the day the engagement ends, what you paid for wasn't a solution. It was a leash.

What a good answer sounds like: They talk about handoff before you even ask—documentation, training, a plan for what your team owns once they're gone. A consultant who's a little uncomfortable talking about their own exit is telling you something true.

7. Tell Me About an AI Project That Failed

Watch their face as much as their answer.

Why it matters: Everyone who's done real work in this space has a failure story. If they don't have one, they either haven't done enough of this work to hit failure yet, or they're not being straight with you about the ones they have.

What a good answer sounds like: Specific, and a little humbling. "I recommended a tool for a client that looked perfect on paper and nobody on their team ever opened it. I learned to test adoption before I recommend anything now." That's a person who's actually done the work.

8. Why Shouldn't I Just Do This Myself?

Save this one for last, and pay close attention to how they handle it.

Why it matters: An honest consultant will sometimes say: you should. If every answer to this question loops back to why you need them, that's not honesty. That's a sales script wearing an honesty costume.

What a good answer sounds like: A real, sometimes uncomfortable answer. "If you've got a few hours a week and one clear workflow, you might not need me yet—start there and see how far you get." We laid out that whole decision on its own in do I need an AI consultant, because the honest answer depends entirely on your situation, not on who's selling you the call.

Questions to Ask an AI Consultant: The Red Flags Worth Walking Away From

A few patterns show up often enough that they're worth naming plainly:

• Retainer-first, diagnosis-never. If the pitch jumps straight to a monthly contract before anyone's looked at your actual workflows, you're buying a payment plan, not a plan. We wrote about what AI consulting actually costs and which engagement type fits which situation—retainers included, but only once the diagnosis is done.
• Tool-pushing without context. The same three tools, every client, regardless of what your business actually does.
• Jargon with no plain-language follow-up. If "AI transformation" and "intelligent automation" never turn into an actual sentence you could explain to your team, that's decoration, not a strategy.
• Guaranteed outcomes. Nobody can promise you a specific ROI from a tool they haven't tested against your real workflows yet. Confidence is fine. Certainty is a sign they're selling, not thinking.

None of these are complicated to spot once you're looking for them. That's exactly why they work on people who aren't.

One Step Before You Sign Anything

Eight questions. That's not a lot to ask of someone who's about to get real access to how your business runs.

Bring these to any consultant you're considering—including me. Ask them straight, and listen for whether the answers are specific or slippery. That's the whole point of writing this list down: not to sell you on one consultant over another, but to make sure whoever you choose has actually earned the fee before you pay it.

You know your business better than any consultant walking in the door. I just want you asking these eight questions before you hand anyone the keys to it.

If you'd rather ask them to someone directly, on a call, that's exactly what our AI consulting engagements are built to answer first—before we ever talk about a retainer.

Which one of these eight would the last consultant you talked to have struggled to answer? Start there. Ask it again, out loud, before you sign anything.