Most organisations we meet are not short of AI ideas. What they lack is a way to tell which ideas are real. Somebody has a list, perhaps predictive maintenance, a document assistant and quality inspection from photographs, and every vendor who sees it says yes to all of it.
We say no to some of it, and we back that up with a measurement. If you are looking for an AI strategy, this is a version of one that starts from what your data can support.
Two ways in
Readiness Scan
€4,950 excl. VAT. Fixed price, typically two to four weeks. The scan combines one workshop day with the people who own the decisions and a structured review of the data and systems you already have. It produces a scored, prioritised list of up to eight candidate use cases, in which the ones that will not work are marked as such, with the reason. Nothing runs on your data at this stage, which is what keeps the scan fast. It ends in a written report and a one-hour readout.
Feasibility Assessment
€24,500 excl. VAT. Fixed price, typically eight to twelve weeks. This is the measured version. We take the two or three most promising candidates and run real baselines on your actual data to establish what accuracy is achievable and with what uncertainty, where the data is not yet good enough, and roughly what deploying the result would look like and cost to run. You get numbers measured on your own data, along with a build outline you could hand to any supplier.
The scan tells you where to look, and the assessment tells you what is there. Most clients start with the scan, partly because its data review is also the check that makes a fixed price for the assessment possible. Some already know which question matters and go straight to the measurement.
Timelines start once the data and access are in place, not at signature. If the data turns out not to support the question, we write that up as a finding, stating what is missing and how much would be enough, so it does not turn into a stalled project.
What you get
- 1
Decision-maker workshop
We spend a day with the people who will sign off. Which decisions would a model change, what does being wrong cost, and what do you already measure?
- 2
Data and systems review
We establish what exists, where it lives, how clean it is and who owns it. Most infeasibility is discovered here, before any modelling.
- 3
Scored shortlist
Each candidate is rated on feasibility, data readiness, value and risk, including the ones we recommend dropping, with the reason.
- 4
Baselines on your data (assessment tier)
We fit simple, fair baseline models on the top candidates to show the achievable accuracy, the calibration, and what is missing.
- 5
Readout and roadmap
The results are presented to the same people who were in the workshop, with a prioritised plan you can defend and what it would take to start.
Not included: picking vendors or tools for you, extracting or labelling your data, and building the solution. If the answer is to build, we quote that separately, and you are free to take the plan elsewhere. How your data is handled during the work is your choice at intake; the three levels are set out with the other assessments.
Who this is for
The assessment is meant for leadership who need to decide where AI belongs before committing a budget, and for R&D groups with more ideas than capacity. It also suits anyone who has been told "AI can do that" and wants a second opinion from people who build the models rather than sell the software.
It works equally well as a first engagement with us or as an independent check on a plan someone else proposed.