The Questions to Ask an AI Consultant Before You Hire One

I am one of these, so read this with that in mind. I am also the person who has watched small businesses hand over four figures for a slide deck, and I would rather you asked hard questions of me than got burned by somebody else.

Here is what I would ask.

What will I actually have at the end of this?

The single best question, and the one that separates the field.

A good answer names a thing. A tool that does X. A system that handles Y. Something you can point at. A bad answer describes a process: a discovery phase, a roadmap, a strategy, a framework.

Roadmaps are not worthless, but you can usually get one from a conversation. If somebody wants several thousand pounds to produce a document describing what could be built, ask what it costs to just build the first thing instead.

Who is doing the work?

At a consultancy, the person selling to you is frequently not the person delivering. That is not dishonest, it is how the model is priced, but you should know before you sign whether the senior person you are talking to will be on the job or on the account.

With an independent it is one person, and you are looking at them.

What happens when it gets something wrong?

Everything built on AI gets things wrong sometimes. Anyone who tells you otherwise has not run one in production.

The question is what happens next. Does it fail quietly and you find out from a customer, or does it flag and stop? Where does the exception go? Who looks at it?

If they have not thought about this, they have not shipped much.

Can I run it without you?

This is the one people forget to ask and regret later.

Some builds hand you something you own and can operate. Others leave you dependent on the person who made it, sometimes deliberately. Ask directly: if we stopped working together in six months, what happens to this?

A straight answer is a good sign. Evasion is a very good sign, of the wrong kind.

What is the ongoing cost?

Two parts. The software it runs on, which for most small business builds is somewhere between ten and eighty pounds a month. And the person, if you need one.

Be careful with monthly retainers attached to work that does not need maintaining. A tool that runs a fixed job does not require someone to tend it every month. If a retainer is proposed, ask what happens in month four when the build is done.

Have you built something like this before, and what happened?

Not "do you have experience". Ask for a specific thing they built, for a specific business, and what changed as a result.

Watch for numbers. Watch for whether they can describe the awkward parts. Anyone who only tells you about the smooth ones is either lucky or editing.

What would you tell me not to do?

The most revealing question on the list.

Anyone genuinely on your side has a list of things they would talk you out of. If everything you suggest gets an enthusiastic yes, you are talking to a salesperson.

I will tell people to spend twenty pounds a month on an assistant subscription and give it an afternoon rather than pay me. That has cost me work. It has also meant the people who do hire me stay.

How do you charge, and what is it based on?

Fixed price for defined work is usually best for a small business, because you know the number before you commit and there is no incentive to take longer.

Day rates are fine if the scope is genuinely unclear, but ask how many days and what happens if it runs over. Hourly billing on a small build is where budgets quietly disappear.

For context on the market, what AI automation actually costs has real numbers.

Why me, and not the cheaper option?

Ask it plainly. If they cannot explain what you get for the difference, there may not be one.

There are good answers. Being the person who does the work rather than passing it down. Building on tools you already own instead of forcing a switch. Being around afterwards. Being reachable on a Tuesday when something breaks.

There are also bad answers, which mostly sound like adjectives.

The one that matters most

If none of that sticks, keep this: ask what changes on a Tuesday.

Describe your ordinary Tuesday afternoon with the thing working. If the answer is "you would be using AI", walk away. If it is "you would not spend two hours writing quotes", that is a project worth paying for.

If you are still deciding whether to hire anyone at all, is an AI consultant worth it is the same honesty applied to hiring me, and agency, freelancer or do it yourself covers which route suits.

Ask me all nine. I would rather answer them than have you find out later. Get in touch, or see what I build.