What to look for in an AI consultant
If you're looking for an AI consultant, or trying to figure out who counts as an AI expert, you're probably not short on options. Plenty of people will take the call. The hard part is telling the difference before you sign something.
I get asked about this as Co-founder of AI Answered, so I have a bias. I'll say that up front. The checks below are the ones I'd want you to use on anyone, including us.
They should start with your business, not their tool.
A useful first conversation is about your week. Which tasks repeat? Where do things wait or go wrong? Where does the money come from? What does a better result look like, in a number you already track?
If the call jumps to a platform before they've asked how the work runs, they're fitting your company to what they sell.
The same is true of titles. "AI expert" doesn't mean much by itself. What you want is someone who can explain, in plain language, which part of your business they would touch first, and how you'd know it worked.
If they can't make the complex simple in the first meeting, it's unlikely they'll do it in the work.
Ask to see how they think, not a highlight reel.
Portfolios full of logos and the word "transformation" tell you almost nothing. Ask what problem they were solving, how they measured it, what went wrong, and who runs the system now.
You don't need a dramatic story. You need evidence they can talk about the work after the demo ended.
Be careful with anyone who can't tell you what happens when the model is wrong. It will be wrong sometimes. A serious partner will talk about review, logs, limits, and a way to turn the thing off. "It doesn't make mistakes" is a reason to end the call.
Look for a small, honest first project.
A good partner will want a short first engagement with a decision at the end: keep going, change course, or stop. They'll want a baseline and a target before they build.
Vague phases with no dates are a warning. So is a plan to "bring AI into the whole company" in one move. That's not a plan. That's a slogan.
Also ask who does the work. If the person on the sales call is not the person you'll work with, meet the person you'll work with before you agree to anything.
Ask what they need from you each week. Access to systems, a person who knows the workflow, time for feedback. A consultant who says they need nothing from you has not thought it through.
Make sure you keep what gets built.
Ask where the system will live, who owns the accounts, and whether your team can run it if the relationship ends. You should own the workflows, prompts, documentation, and access.
If everything lives in their login, and nobody on your side can explain it, you didn't hire a partner. You rented a dependency.
Ask about data too. Where it goes, who can see it, and whether a model provider trains on what you put in. If you handle customer, financial, or other sensitive information, that conversation should happen early, not after the build.
This is also where simple rules inside your own company help. Which tools are approved, what data may not leave your systems, and who checks an output before it goes out. A partner should be comfortable working inside those rules, or helping you write them. They should not ask you to ignore them.
Pricing should follow the value, not a menu.
I won't quote prices here, because the number should come from the business in front of you. What I would listen for is this: do they want to understand your numbers before they talk about fees? Can they tell you what you are paying for, and how you'll judge the work?
A packaged plan that doesn't look at your company first is a product. It may still be useful. It is not the same thing as a partner.
At AI Answered we work with established companies as a growth partner. We look at the numbers first, then price on the value we can deliver. That isn't a line I added for this post. It's how I'd want to be hired, and how I'd hire someone else.
A short list you can take into the call.
You can get most of the way there with a handful of questions:
What part of our work would you start with, and why?
How will we know it worked? What's the number, and what's the baseline?
What happens when the AI gets something wrong?
Where does our data go, and who owns what you build?
Who will actually do the work, and what do you need from us each week?
If those answers are clear, you're in a much better place than a demo can put you.
If you want to see how we work, I wrote it up on the AI Answered growth partners page. We start with a review of your numbers. If we can't help, we should say so.
The first step, before you hire anyone, is where to start with AI in your business. If the company hasn't measured itself, AI won't fix a business that doesn't know its numbers.