Choosing an AI consultant: what to look for

What to look for in an AI consultant

Plenty of people and agencies offer AI advice. You only see the differences when you ask follow-up questions. These eight points help, each with a question for the first meeting.

1. 1

   The conversation starts with your work, not a tool

   A consultant who recommends a platform or model in the first meeting hasn't seen your processes yet. Good advice starts with where time, money or revenue is being lost.

   Only then comes the question of which technology fits. Sometimes that's AI, sometimes an integration between two systems, sometimes a setting in software you already have.

   Ask in the meetingWhat do you want to know about us before you recommend anything?
2. 2

   You get a figure you can check

   Advice without a figure is hard to decide on. Ask how the consultant works out what the current work costs: the hours, the hourly cost, fixing errors and work that gets left undone.

   A figure you can check yourself is a figure you can explain internally.

   Ask in the meetingHow do you calculate what our current work costs, and can we follow every step?
3. 3

   They also tell you where AI doesn't fit

   Not every process gets better with AI. Work with fixed rules is often better served by regular automation, and a process that runs on habit needs sorting out first.

   A consultant who never says "not here" is mainly advising what they can sell.

   Ask in the meetingWhich of our ideas would you not use AI for, and why?
4. 4

   The advice can be carried out

   A forty-page strategy document ends up in a drawer. A short list is more useful: which opportunities first, what they need, who does what and what the first step costs.

   Ask in the meetingWhat will we have in hand when you're done, and could someone else carry it out?
5. 5

   They look at your systems and your data

   Many AI ideas get stuck on the basics: systems that can't be connected, data spread across six places, or nobody who owns it. If nobody looks at that, the plan won't fit in practice.

   Ask in the meetingWhich systems and data do you want to see before you give advice?
6. 6

   It's clear who carries it out afterwards

   Advice and delivery can sit with the same party or with two different ones. Both can work, as long as it's clear who is responsible if the plan turns out differently in practice.

   If advice and delivery are separate, ask how the handover works and who your point of contact is after go-live.

   Ask in the meetingWho carries out the plan, and who do we talk to if it doesn't work as intended?
7. 7

   Privacy and regulation are part of the advice

   AI raises questions about personal data, access to systems and the EU AI Act. Those belong in the advice, not after something has already been built.

   Ask, for each opportunity, which data goes in, to which supplier, and what needs to be arranged.

   Ask in the meetingHow do you factor the GDPR and the AI Act into your advice?
8. 8

   The first step has a fixed price and an end date

   An advisory engagement without an end date gets expensive without you knowing why. Ask for a defined first step, with a fixed price, a lead time and a deliverable you know in advance.

   Ask in the meetingWhat does the first step cost, how long does it take and what exactly do you deliver?

Questions to ask in the first meeting

Take this list to your first meetings. A partner that suits you gives a concrete answer to every question, even if that answer is that something can't be done.

What does an AI consultant do?

An AI consultant helps an organisation decide where AI or automation adds value and how to get there. That can be a short exploration or a full strategy engagement. The title isn't protected: it can be an independent adviser, a large consultancy or a company that also builds.

An AI adviser does the same work in practice. The difference isn't the title, but what you have in hand after the advice and who carries it out.

What belongs in an AI strategy for an SME?

For an organisation of ten to a hundred and fifty people, an AI strategy isn't a thick document. It's a short set of choices your team understands and can carry out.

• What the current work costs, in one figure you can check.
• The three opportunities with the highest value, and why those.
• What you deliberately won't do.
• What needs to be in place first: access to systems, data and ownership.
• The first process, with a plan for the first 90 days.
• Agreements on which AI tools are allowed and which data may go into them.

Everything else can wait. A strategy you adjust in three months is better than one you never carry out.

Signs a consultant isn't right for you

• A tool or platform is on the table before anyone has seen your processes.
• Nobody can explain how the figure in the advice was built up.
• Every process turns out to be suitable for AI.
• The deliverable is a presentation, without a first step or a price.
• After the advice, it's unclear who carries it out.
• The engagement has no end date.

Frequently asked questions