How Much Does an AI Consultant Cost? A Transparent Guide
The honest answer is that price tells you almost nothing, because “AI consulting” covers five completely different things. A self-serve course and an enterprise strategy engagement are both sold under that label, and they share nothing except the words on the invoice.
The number that matters is not the fee. It is what your team can do differently the month after everyone goes home — and that varies far more than price does. This guide explains what each format actually delivers, what drives the cost up, and the four questions that tell you more than any quote will.
The five formats, and what each really delivers
Almost every offer on the market is one of these five. They are listed cheapest first, and the pattern worth noticing is that cost and adoption do not move together in the way you would expect — the most expensive format is not the one most likely to change how people work.
| Format | What you get | What you don’t get | Adoption a month later |
|---|---|---|---|
| Self-serve course | Video lessons, exercises, sometimes a certificate | Any connection to your actual work | Very low — most corporate course licences are never completed |
| Lecture-style training | A trainer explaining concepts and demonstrating tools | Anything built for a specific person’s job | Low — people leave informed but with nothing to use on Monday |
| Hands-on workshop | Every participant builds something for their own role | Support once the room empties, at the cheaper end | Good, and much better when support follows |
| Ongoing partnership | Continued support, new use cases, someone to ask | The initial shift — this compounds change, it doesn’t start it | Highest, because it covers the weeks when habits form or die |
| Strategy consulting | Discovery, maturity assessment, a roadmap, executive sessions | Any guarantee that a single person works differently | Often near zero — it is optimised to produce a document |
Strategy consulting is by far the most expensive line and reliably the weakest on adoption. That is not because the consultants are bad. It is because the engagement is designed to answer “what should we do about AI?” when the actual problem in most mid-size companies is that nobody has changed how they work. A roadmap does not close that gap, and it is an expensive way to discover it.
What actually drives the price
When two quotes differ substantially, it is almost never the session itself. It is these four things, and they are worth asking about directly because each one is the difference between a good day and a lasting change.
- Preparation before the day. Whether someone mapped your real workflows in advance, or whether you are getting the same session every client gets. This is the single biggest quality difference and it is invisible in a proposal unless you ask.
- Role-specific building versus general demonstration. Everyone leaving with something built for their own job costs more to run than one person presenting to a room. It is also the entire reason the session works.
- What happens afterwards. The weeks after the session are when new habits either stick or quietly die. Support through that window costs money and is the difference between a spike and a change.
- Whether anyone measures. Providers who track what people actually do afterwards charge more, because they have to be right. Providers who measure satisfaction instead are selling you a pleasant day.
How to work out if it is worth it
You will find no shortage of ROI calculators that multiply hours saved by an hourly rate and produce an enormous annual figure. I would not put much weight on them, and we do not publish one, for a straightforward reason: every one of those models assumes adoption, and adoption is the only variable that actually matters.
Run the arithmetic yourself and the assumption becomes obvious. Take the hours your team spends on work AI genuinely compresses — writing, summarising, researching, formatting, drafting, processing. For most knowledge teams that is a substantial share of the week. Now assume half of it comes back. The resulting number is always large, which is exactly why it is not useful: it is the same large number whether the engagement works or not.
The number that separates a good investment from a bad one is what proportion of your team is still working the new way a month later. A workshop everybody enjoyed and nobody applied has negative return — you paid for it, and you also spent some of your organisation's willingness to try again. So evaluate the methodology, not the arithmetic.
Four questions to ask any provider
Ask these before you compare a single number. The answers are more predictive of value than any quote, and vague answers to any of them tell you what you need to know.
- What will each person have built by the time they leave the room? If the answer describes what people will have learned rather than built, you are buying a lecture.
- What happens in the month after? This is when adoption dies. A provider with no answer has not designed for the thing that determines whether any of it lasts.
- How do you prepare, and what do you need from us beforehand? “Nothing” means you are getting the standard session.
- What do you measure, and can you tell me what it showed last time? Hours saved and use cases adopted are behavioural. Satisfaction scores are not.
How we price, and why
We quote a fixed price, agreed before you commit to anything, and we never bill by the hour. That is a deliberate choice rather than an administrative one: hourly billing rewards us for taking longer, and everything about this work should reward us for making the session land and then getting out of the way.
The Kickstart is a half-day — around three hours — and the price is the same whether your team is twenty people or a hundred and fifty. We will give you the figure in the first conversation. We do not publish it because it varies by sector and geography, and a published number that is wrong for you is worse than no number at all.
The Partner programme that follows is a monthly fee you can stop whenever your team no longer needs it. The intention is genuinely for it to end. We are trying to build a team that finds its own use cases without us, and when a client tells us they are there, that is the result we were aiming at — not a lost account.
If you want the number for your team, the fastest route is a conversation — I take them myself. The Deployed Kickstart is the half-day session; the Partner programme is what keeps it from fading.
Frequently asked questions
How much does an AI consultant cost?
It varies enormously, because “AI consulting” covers five different things — self-serve courses, lecture-style training, hands-on workshops, ongoing partnership, and strategy consulting. They share a label and almost nothing else. Price tells you very little on its own; what separates a good investment from a bad one is what proportion of your team is still working the new way a month later.
Is an AI consultant worth the money?
It depends entirely on whether the engagement is designed around behaviour change or around producing a document. A workshop everybody enjoyed and nobody applied has negative return — you paid for it, and you also spent some of your organisation’s willingness to try again. Evaluate the methodology rather than the arithmetic, because every ROI model assumes adoption, and adoption is the only variable that actually matters.
What makes some AI consulting more expensive than others?
Four things: whether anyone mapped your real workflows before the session, whether every participant builds something for their own role or watches a demonstration, whether support continues through the weeks when habits form or die, and whether the provider measures what people actually do afterwards rather than how satisfied they felt.
What should I ask an AI consulting provider before buying?
Ask what each person will have built by the time they leave the room, what happens in the month afterwards, how they prepare and what they need from you beforehand, and what they measure. If preparation requires nothing from you, you are getting the standard session. If they measure satisfaction rather than behaviour, they are selling a pleasant day.
Why is AI strategy consulting often poor value for mid-size companies?
Because it is designed to answer “what should we do about AI?” when the actual problem is usually that nobody has changed how they work. It is the most expensive format and reliably the weakest on adoption — not because the consultants are poor, but because a roadmap does not close that gap, and it is an expensive way to find that out.
How does Deployed price its services?
A fixed price agreed before you commit, and never hourly — hourly billing rewards taking longer, and this work should reward making the session land and then getting out of the way. The Kickstart price is the same whether your team is twenty people or a hundred and fifty. The Partner programme that follows is a monthly fee you can stop whenever your team no longer needs it.
Found this useful? Send it to someone who needs it.