How to choose an AI consultant: what they do, what it costs and the questions to ask before you hire

As soon as you start looking for an AI consultant, you notice that everyone says roughly the same thing: "we help you with AI", "we build tailored solutions", "we free up time with automation". The words are the same whether the person behind them has been building production systems for ten years or saw their first API last quarter. This guide shows how to tell the difference before you sign – and what you are actually buying when you hire an AI consultant.

If you are still unsure whether you are ready to bring in outside help at all, start with Is your company ready for AI?. This article assumes the decision has been made and you now need to choose the right person or partner.

What does an AI consultant do – and what do they not do?

An AI consultant translates business problems into AI solutions. That is the whole point. The job starts not with the technology but with your business: where time leaks away, where decisions are made on gut feeling, where people spend their days moving information between systems by hand. Only then does the question arise of whether AI is the right tool – and a good consultant says no as often as yes.

That is the most important quality to look for. Anyone who sells AI as the answer to everything has stopped listening to the problem. A good consultant advises against AI when a simple integration, a clearer process or an off-the-shelf tool solves the same thing more cheaply and more reliably.

What a good AI consultant typically delivers:

• Analysis of the current situation and prioritisation – mapping where in the business AI makes a real difference, and ranking opportunities by value against effort.
• Choice of technology and supplier – which model, which platform, build or buy. An area where the wrong choice is expensive to reverse.
• Implementation – actually building: automating manual workflows, AI agents that carry out tasks, or RAG solutions that let a model answer from your own data instead of general knowledge.
• Maintenance – follow-up, adjustment and operation.

What an AI consultant is not: a supplier of magic. No serious consultant promises that AI will replace a department or double your margin without first having seen how you work. If that kind of result is promised before anyone has understood your business, it is a warning sign, not a selling point.

When does your company need an AI consultant – and when is an off-the-shelf tool enough?

Not every AI need calls for a consultant – far from it. If you just want the team to write faster or summarise meetings, an off-the-shelf tool is enough – a licence and a short internal training session will do. A consultant only becomes worth the money when the need is specific to your business and no off-the-shelf product fits as it is.

Signs that it is time to bring in help:

• Much of the working day goes on manual information work – copying between systems, searching in documents, piecing reports together.
• You know AI ought to be able to help, but not where to start or what is realistic.
• Earlier AI attempts got stuck at the experimental stage and never made it into production.
• You need someone who can honestly assess whether you are ready at all – before you invest in a build.

The last point is underrated. One of the most valuable things a consultant does is advise against something: to say "you should fix this in your processes or your data structure first, otherwise you are building on sand". That saves more money than any build.

AI consultant, AI agency or in-house skills – what suits you?

There are three routes to AI expertise, and they are not mutually exclusive.

An AI consultant is usually an individual or a small team who digs deep into your business, sets priorities and points the way. The strengths are advice, independence and one person taking responsibility for the whole. This suits you when you need direction and decisions more than pure build capacity.

An AI agency has more delivery capacity – several people, more skills, the ability to build and maintain larger solutions over time. It is the right choice when you know what you want built and need the muscle to do it.

In-house skills – hiring or upskilling – are right when AI becomes an ongoing, central part of the business and the need is permanent. The drawback is time: recruiting and building up know-how takes months, and a bad hire is expensive.

In practice, many companies combine them: a consultant to set the direction and confirm you are ready, an agency or a small team to build, and in-house staff to take over operations once the solution is established. A good external partner should actually help you get there – making itself less necessary over time, not more.

What does an AI consultant cost?

The honest answer is that it depends on the nature of the assignment, and that anyone who gives a fixed price before understanding your problem is guessing. Pricing usually follows one of three models.

• Hourly rate – common for advice, workshops and smaller investigations. Good when the scope is unclear and you want to keep the risk down early on.
• Fixed project price – for a clearly defined delivery with a known scope, for example a specific build. This requires both sides to be able to describe the end result in advance.
• Retainer – an ongoing monthly arrangement for continuous development and maintenance. It suits you when AI becomes an ongoing part of the business rather than a one-off project.

What matters most is not the model but that the price reflects the business value. An automation that saves twenty hours a week is worth something completely different from one that saves two – and the cost should reflect that, not just the number of hours worked. A good consultant talks about value and payback, not just their own hourly rate.

The questions to ask before you hire an AI consultant

This is the heart of it. Ask these questions in the first conversation – the answers quickly reveal who understands your business and who is selling technology for its own sake.

1. Which of my problems would you not solve with AI? Anyone who cannot answer sees AI as the answer to everything. A good answer shows judgement.
2. Can you show something you have actually built and put into production? The difference between a demo and a system that has actually run live is enormous. Ask for concrete examples.
3. How will we measure whether this has worked? A serious consultant defines outcomes in advance – time saved, fewer errors, faster answers – not "we'll see how it goes".
4. What happens to the solution when the assignment ends? Do you own the code and the data? Can you maintain it without the consultant? Avoid arrangements that leave you locked in.
5. Where will my data come from, and where will it go? Particularly important for RAG solutions and anything involving sensitive information. The answer should be concrete, not vague.
6. What is required of us internally for this to work? Every real AI project requires something from the client – data, time, access to systems. Anyone who says "nothing" has not thought it through.
7. How comfortable are you with challenging me? Ask it outright. A consultant who never advises a client against anything is optimising for billable hours, not for your business.
8. Who maintains it afterwards – and what does it cost? AI solutions need ongoing adjustment. An arrangement that ignores maintenance ignores half the picture.

You do not need to ask all eight. But if the answers to the first three feel vague, you have already learned the most important thing.

How to identify a serious AI consultant

The clearest divide in the market runs between two kinds of supplier. One knows the technology but does not understand the business – they build impressive things that solve the wrong problem. The other talks business and impact but lacks a technical foundation – they sell a vision they cannot build. What really makes the difference is the combination: an understanding of what drives your business, and the genuine technical knowledge to build something that holds up in production and over time.

Concrete signs that you are talking to the right person:

• Asks more questions about your business than about which AI model you want to use.
• Advises against things – and can explain why.
• Shows concretely what they have built, not just what they can do in theory.
• Talks about ownership, maintenance and what happens after launch.
• Relates price to value and payback, not just to hours.

The buzzword seller does the opposite: leads with the technology, promises results before understanding the problem, is vague about what has actually been built and goes quiet when you ask about ownership and operation. The signals are not subtle once you know what to listen for.

Next step

If you are bringing in AI help, what matters is less which technology is hottest and more who understands your business well enough to choose the right technology for you – and is honest enough to advise against it when needed. If you want to hire someone who combines business understanding with technical depth, or just want to talk through whether AI is the right route for you, you are welcome to get in touch. For deeper automations, AI agents and RAG solutions on your own data, we work closely with our sister brand AI Partner.