What Clients Expect vs What Actually Happens

You probably know how this is supposed to go. You book a call with an AI consultant, and somewhere in your head you are already braced for the deck: the market-trends slide, the maturity-curve graphic, the logo wall of clients, and a closing slide with three pricing tiers helpfully labelled Good, Better and Transformational. So when people ask me what to expect from an AI consultant, my honest answer is that the first conversation should feel like the opposite of that. It should feel like two engineers trying to work out whether a problem is real.

I have sat on both sides of these meetings. Here is the gap between what most clients expect on that first call and what actually happens on ours — and why the difference matters more than it sounds.

The pitch you have been trained to expect

There is a reason you walk in defensive. The standard consultancy sales call is engineered to do three things: establish authority, manufacture urgency, and get you to commit to a large scope before anyone has proven anything works. The vocabulary is a tell. You will hear about "digital transformation journeys" and "AI-first strategies" and a "phased enablement roadmap". What you will rarely hear is a straight answer to a simple question: will this actually save you money, and how will we know?

None of that is malicious. It is just what happens when the people selling the work are not the people who have to build it. Slides are cheap. A working system is not. And when the sales motion is disconnected from the engineering reality, the incentive is to sell as much scope as possible and worry about delivery later. You end up paying for confidence, not competence.

What actually happens on the first call

Our first conversation is mostly me asking awkward questions. Not "what's your AI strategy" — that question has no useful answer. Instead: what is the specific, repetitive, expensive thing your team does every week that you wish they didn't? Where does the work pile up? Who is doing manual copy-paste between two systems at 6pm? What happens today when that process breaks?

I am not trying to impress you. I am trying to find a use case small enough to prove and valuable enough to matter. Often we find it in the first twenty minutes. Sometimes we find that you do not need AI at all — that a database query, a cleaner form, or fixing one broken integration would solve the thing you were about to spend twenty thousand euros on. I would genuinely rather tell you that than sell you a model you don't need. That is the part most clients are not expecting, and it is the part that tends to earn the most trust.

A quick example of how this plays out. A founder came to us convinced they needed a custom chatbot trained on their product catalogue. Twenty minutes of questions later, the real problem turned out to be that their support team was answering the same five questions over and over because those answers were buried three clicks deep on the website. The first fix was not a model at all — it was moving those answers to the top of the page. The chatbot came later, scoped to the genuinely ambiguous questions, and it worked precisely because we had already stripped out the noise. Cheaper, faster, and far easier to maintain than the thing they walked in asking for.

If you want the longer version of how I run these conversations, I have written about what the first ten minutes of a call reveal and why I end most discovery calls in under thirty minutes. I won't repeat all of it here. The short version: a focused call respects your time more than a long one.

The expectations we reset, gently

A few assumptions come up on almost every call. None of them are silly — they are exactly what the market has trained people to expect. But they tend to point you in the wrong direction.

"Where is the big strategy document?"

You might expect to leave with a forty-page strategy deck. We would rather you leave with one clearly defined problem and a rough sense of what proving it would cost. A document nobody reads is not a deliverable; it is an invoice with a cover page. Strategy matters, but it earns its place after you have shipped something real, not before.

"When is the big launch?"

The expectation is a single dramatic go-live. The reality is that good AI work is incremental. We aim for a working prototype you can actually click — usually by the second call — and then we improve it against real usage. A production AI implementation that survives contact with your real data is worth more than a launch event that demos beautifully and falls over in week two.

"How long is the contract?"

People often expect to be locked into a twelve-month engagement before anything ships. We scope the opposite way. An audit from around €2,500 maps the right use case. A proof of concept from around €20,000 proves it works on your data. Only once it works do you decide whether to scale to production. You should never be paying for a year of faith.

Why we work this way

It is not a marketing position. It is the only honest way to sell something that genuinely might not work. AI projects fail for boring reasons far more often than exciting ones — messy data, a process nobody actually agreed on, a use case that looked great in a slide and made no sense in production. The fastest way to find those failure modes is to build a small, real thing early, not to plan a large, imaginary thing thoroughly.

There is also a compliance dimension that rewards this restraint. The EU AI Act is phasing in its obligations through 2026 and 2027, and the systems that will be easiest to govern are the narrow, well-understood ones — not the sprawling "AI everywhere" programmes. Scoping tightly is not just cheaper. It keeps you on the right side of the rules you will have to answer to later. (This is general information, not legal advice; check the European Commission's own guidance for your situation.)

What we expect from you

Honesty cuts both ways. The first conversation works best when you bring the unglamorous truth: the spreadsheet held together with macros, the process that only one person understands, the data that is messier than the org chart admits. We cannot scope around problems you hide. The clients who get the most out of working with us are the ones who treat the first call as a diagnosis, not a performance.

So here is the simplest way to set your expectations before we talk. You are not going to be pitched. You are going to be asked a lot of specific questions, you may be told your problem is smaller or different than you thought, and you will leave with a clearer picture of whether AI is worth your money here at all.

If that sounds more useful than another polished deck, review our transparent pricing or book a free consultation and we will map your first use case together — and tell you honestly if you don't need us yet. You can also browse how this played out for other clients to see what "small and real" turns into over time.

Frequently asked questions