A shortlist of things worth doing, sized — and the ones to leave alone.
Most companies do not need an AI strategy document. They need to know which two or three things to do first, what each is worth, what it will cost, and what could go wrong.
We are the ones who would build it, so the estimates are ours to stand behind. That tends to make the advice more careful.
- Where would AI actually take work off our people?
- What is each opportunity worth, and what does it cost to build?
- Is our data good enough to do this at all?
- What are we allowed to do under GDPR and the EU AI Act?
- Build, buy, or leave it for now?
Advice from people who ship
A strategy written by people who have never run an agent in production tends to underestimate the boring parts: integration, data quality, adoption, the cost of being wrong at scale.
We build and operate our own AI products. When we size an opportunity, we are sizing work we know how to do.
Where work is repetitive, high-volume and judgment-light — the places AI pays — assessed against your operations, not a generic industry list.
Value, build cost, run cost and payback per opportunity, at a level a CFO can challenge.
An honest read on whether the data exists, is accessible, and is good enough — before anyone commits to a build.
What GDPR and the EU AI Act mean for each candidate, including risk classification and where consent or oversight is required.
Where a product already does the job well, we will say so — including when the answer is not us.
Which one first, and why. A programme nobody can fund is worth less than one change delivered this quarter.
Two to four weeks, then a decision
- 01Interviews
The people running the operation, not only the people sponsoring the project. Where does the time go, and what breaks most often.
- 02Data & systems review
What you hold, where it sits, what state it is in, and what the integration effort realistically looks like.
- 03Sizing
Each candidate costed and valued, with the assumptions written down so you can argue with them.
- 04Recommendation
A ranked shortlist, a first project with a scope and a price, and the list of things we advise against.
Written findings you can take to a board — not a slide deck of AI trends.
Value and cost per opportunity, with the assumptions visible and challengeable.
The advisory engagement stands alone. If you take it to another firm to build, that is a legitimate outcome.
Sometimes the recommendation is to do nothing yet
If the data is not there, if the process needs fixing first, or if the value does not clear the cost, we will say so and explain why. That answer is cheaper than a pilot that quietly dies.
The most common finding. Automating a broken workflow makes it break faster.
Where a mature product already solves it, building is an expensive way to be second-best.
Sometimes the first project is getting the data into a state where AI is even possible.
Where advisory usually leads
Bring us the AI question your board keeps asking.
A discovery session is a working conversation, not a demo. You will leave it with a clearer view of what is worth doing — including if the answer is nothing yet.