How do you choose an AI consultant?
Ask every candidate the same ten questions and disqualify anyone who dodges more than two. The questions below test the things that predict whether you get a working system or a deck: pricing candor, incentive alignment, proof, and what happens to your data.
They are drawn from the same standard we publish against ourselves — the “what we don’t promise” list on our method page. Print this page and bring it to every meeting, ours included. If our answers and this list ever disagree, that is your red flag.
The ten questions
Each comes with what a good answer contains — and what a bad one sounds like.
Run the meeting the same way every time: same questions, same order, answers written down. Comparing firms only works when the inputs match, and taking notes changes the answers you get — vague firms get specific or get visibly uncomfortable, and both are information.
1. What will this cost, and when do I learn the exact number?
A builder gives you a range now and a fixed price in writing before work begins. Anyone who needs three discovery calls to produce a number is already billing you; the meetings are the meter. Our ranges are published — $3,500–$8,500 for an audit, $12,000–$45,000 for a sprint — and the whole market’s ranges are in Guide 01, so you will know if a quote is an outlier.
Bad answer: “Every engagement is unique — let’s schedule a discovery workshop.”
2. Will you guarantee an ROI number?
This one is a trap, and the right answer is no. Real results depend on your execution, so a serious firm quotes ranges and shows its assumptions; a firm that guarantees “300% ROI” is quoting fiction with confidence. We publish the assumptions behind our estimates precisely so nobody has to take a promise on faith.
Bad answer: “Clients typically see 10x returns in 90 days, guaranteed.”
3. How long until something actually works?
About 90 days per workflow is what real implementation takes — scoping, building, testing inside your operation, and training your people. Two weeks is a demo, not a deployment, and “transformation by Friday” is the loudest hype tell there is.
Bad answer: “We can have you fully AI-powered in two weeks.”
4. Who scopes the project, and who builds it?
The same person, or you are being bait-and-switched: a partner sells you, then a junior team you have never met inherits you. Ask to name the humans. In our shop the advisor who scopes your project is the one who builds it — that is a published promise, so hold us to it.
Bad answer: “Our senior team oversees all delivery.” (Oversees is doing heavy lifting in that sentence.)
5. Will you tell me if AI is wrong for us right now?
“Wait a quarter” must be an answer the firm can afford to give, or every diagnosis will be “build.” Ask when they last told a prospect not to buy. A firm whose audit can conclude “don’t build yet, and here’s why, in writing” is scoping for your benefit, not their pipeline.
Bad answer: “Every business needs AI today — the risk is waiting.”
6. What tools will you recommend, and what do you earn from them?
Vendor-neutral means the recommendation can be the free tool, and the firm earns nothing when it is not. Reseller commissions and “partner tiers” quietly bend advice toward whatever pays the consultant twice. Ask the question directly and watch for a flinch.
Bad answer: “We’re a certified premier partner of [the thing they are about to recommend].”
7. Where does our data go?
You want three plain-English answers: where the model runs, what it sees before redaction, and who can prove what happened afterward. A vendor who cannot answer those three has answered them. The strongest pattern we can show is MARCUS — a regulated lender’s back office where the models run on the client’s own hardware and PII is stripped before anything is read.
Bad answer: “It’s all encrypted in the cloud, totally secure.”
8. Who can overrule the system?
A named human, with the authority to say stop — or you are buying a machine that buries why decisions get made. Every system worth building drafts the work and routes the judgment to a person who owns it. If the pitch celebrates “fully autonomous, no human in the loop,” the demo is the product.
Bad answer: “The beauty is it runs itself — no oversight needed.”
9. What do we own when you leave?
Everything, and your team trained to run it — or you have signed a lease, not bought a system. Ask what happens on the day the contract ends: who holds the credentials, the documentation, the prompts, the integrations. Dependency is a business model; make sure it is not the one you are buying.
Bad answer: “Most clients stay on our platform long-term.”
10. Show me something you built.
Not slides — a build, with a name on it, and what it does. One verifiable system beats fifty logos on a credentials page. Ours are public: MARCUS, 14 agents running a regulated lender’s back office, and two redacted pages of a real audit deliverable. Ask every firm for their equivalent.
Bad answer: “We’ve delivered AI transformation for leading brands across verticals.”
What are the red flags, in one list?
Guaranteed ROI numbers, transformation measured in weeks, and a price you can only learn through a discovery sequence — any one of these ends the meeting. The rest of the list: scarcity pressure (“two slots left” with nothing verifiable behind it), tool recommendations that always land on a partner product, autonomy sold as a feature instead of a risk, and case studies with no names and no numbers.
Watch for the pilot trap, too: a paid “pilot” that renews into a second pilot, then a third, with production always one phase away. Pilots are legitimate — ours is called a sprint and it ends with a working system in about 90 days. A pilot with no named graduation criteria is a subscription wearing a lab coat.
None of these people are stupid, and some are not even dishonest — hype is a market, and they are serving it. Your job is only to notice you are the product. The Ampersand essay How to Smell the Hype is the longer training course for that nose.
Why publish questions we could fail?
Because we would rather be disqualified by a sharp buyer than hired by a confused one. A client who chose us against a real standard stays; a client who was dazzled churns — so the standard is good business, not virtue.
It is also the standard we already signed. Our method page lists what we promise and what we refuse to promise, and this guide is that list turned into questions anyone can ask any firm. Use it on us first: book the free 30-minute assessment and ask all ten. If you are still comparing, Guide 01 prices the whole market and Guide 03 runs the hire-or-engage math. Once you choose, two companions keep the engagement straight: a scope a vendor can’t inflate, and exactly what an audit should hand you.