Let me be upfront about something: I run an AI consulting firm. That means I have a financial interest in convincing you to hire an AI consultant — ideally me.

So why am I writing a guide that might help you choose someone else, or decide you don't need a consultant at all?

Because that's actually the job. If I point you toward a solution that doesn't fit, we both lose. You waste money and walk away jaded about AI. I get a bad referral and a client I couldn't genuinely help. The only version of this that works long-term is honest advice, even when it's inconvenient.

The AI consulting market is flooded right now. Everyone with a ChatGPT account and a LinkedIn profile is offering "AI transformation services." Some of them are brilliant. Some of them will burn your budget and disappear. The difference isn't always obvious from a sales call.

Here are the seven questions I'd want someone to ask me — and what the answers should and shouldn't sound like.

1. Can you explain what you'd actually build for me — in plain English?

This one is a filter, not a trick question. A good AI consultant should be able to describe their proposed solution in terms a non-technical business owner can follow. Not "we'll leverage a multi-modal LLM pipeline with retrieval-augmented generation" — but "we'll build a tool that reads your incoming support emails and drafts a reply for your team to review before sending."

If they lead with jargon and get defensive when you ask them to simplify, that's a red flag. Jargon is sometimes a shield for shallow understanding. Real expertise tends to make things clearer, not murkier.

Ask them: "If I described this to my bookkeeper, what would I say?" If they can't answer that, keep looking.

2. Do you understand my business, or just AI?

Knowing how to build AI tools is table stakes. The harder skill is knowing when AI actually helps — and understanding enough about your industry, your workflow, and your constraints to make that judgment well.

A consultant who comes in with a pre-packaged solution before they've asked about your team size, your tools, your margins, or your customers is selling something, not solving something.

Before any engagement, a good AI consulting services provider should ask questions like: Where does your team lose the most time? What decisions are you making manually that you wish were faster? What would it mean for your business if this particular bottleneck went away?

If they skip straight to demos and pricing, slow down.

3. What's their track record with businesses your size?

The skills that help a Fortune 500 company deploy AI at scale are genuinely different from the skills that help a 12-person service business automate their intake process. Make sure you're talking to someone whose experience is relevant to your situation.

This isn't about prestige — it's about fit. A consultant who has only worked with enterprise clients may not understand the budget realities, the lean team dynamics, or the "we need this done in three weeks, not six months" pace that small businesses operate at.

Ask specifically: "Have you worked with companies similar to mine? What did you build for them, and what happened after?" Results matter more than logos.

4. Will you build something I can maintain without you?

This is the question most people forget to ask — and it's one of the most important.

There's a version of AI consulting that creates dependency: a custom system that only the consultant can modify, built on proprietary tooling, with no documentation and no training for your team. That's a great recurring revenue model for the consultant. It's not great for you.

A trustworthy consultant builds toward your independence. That means documentation. That means training your team. That means using tools and platforms that have real support communities and don't require a PhD to operate.

Ask them: "What would it take for my team to maintain this after you're done?" If the honest answer is "you'd need to keep me on retainer indefinitely," at least you know what you're getting into. If they look uncomfortable with the question, that tells you something too.

5. How do you measure success?

Vague outcomes are where consulting engagements go to die. "We'll improve your operations" means nothing. "We'll reduce the time your team spends on manual data entry by 40% within 90 days" is something you can evaluate.

Before signing anything, agree on what success looks like — and make sure it's tied to business outcomes you actually care about, not activity metrics like "models trained" or "hours billed."

Good consultants want clear success criteria because it protects them too. It ends the engagement cleanly when the work is done, and it gives them a real result to point to when they're talking to their next client.

If a consultant resists defining specific outcomes, ask yourself why.

6. What do you recommend I not do?

This is the biggest trust signal on this list.

Any consultant can tell you what AI could do for your business. The ones worth hiring will also tell you what AI won't do well, what's not worth the investment at your stage, or what problem you should solve with a spreadsheet and a good hire before you spend $15,000 on automation.

When I meet with a potential client, I try to be honest about the cases where I'd say "not yet" or "not this." It costs me deals sometimes. It also means that when I do say "this is worth doing," people believe me.

Ask a prospective consultant: "What's something businesses like mine often want to do with AI that you think is a mistake?" If they have a real answer — one with nuance and specifics — that's a good sign. If they pivot to telling you everything is possible, be careful.

7. Can I talk to a past client?

References matter. A track record of successful work with real businesses is the most concrete signal you have.

Ask for two or three past clients who are willing to talk — ideally businesses similar to yours in size or industry. When you talk to them, ask the same questions you'd ask the consultant: Did the project stay on budget? Did it deliver what was promised? What was the communication like when things got hard? Would they hire this person again?

If a consultant can't or won't provide references, that's worth noting. New consultants won't have long track records — that's fair — but they should be able to point to something: past work, a case study, a pilot project, something you can evaluate.

Red Flags to Watch For

A few patterns worth keeping an eye out for, regardless of how polished the pitch is:

Guaranteed ROI numbers with no basis. "We'll 10x your productivity" with no data, no case studies, and no clear mechanism is a sales line, not a plan.
Contracts that lock you in before discovery. A legitimate consultant should want to understand your business before proposing a solution. If they're pushing you to sign a six-month retainer on the first call, slow down.
No clear deliverables. If the contract describes "AI strategy consulting" and nothing more concrete, you're buying ambiguity.
Reluctance to document or train your team. As mentioned above, this is a structural conflict of interest. Watch for it.
Overselling the technology, underselling the change management. AI projects fail far more often because of people and process issues than because the technology didn't work. If your consultant isn't talking about adoption, training, and workflow change, they're leaving out the hard part.

One More Thing

Hiring an AI consultant — any AI consultant, including me — should feel like hiring a trusted advisor, not buying a product. You should walk away from the first conversation feeling like you understand your situation better, not more confused or more pressured.

If something feels off, trust that. There are enough good people doing this work that you don't have to settle for someone who makes you uneasy.

If you'd like to have a no-pressure conversation about what AI consulting services might actually look like for your business, I'm happy to talk. We'll figure out together whether it makes sense — and if it doesn't, I'll tell you that too.

No pitch deck. No obligation. Just an honest conversation.

How to Choose an AI Consultant: 7 Questions to Ask Before You Sign Anything