Services
Three fixed checkpoints — days 30, 60 and 90 — that force a decision instead of letting an AI initiative drift. One workflow, one owner, one number.
Most small and mid-sized companies don’t fail at AI adoption because they chose the wrong tool. They fail because nothing forced a decision. Someone raises AI in a leadership meeting, a vendor books a demo, a subscription gets signed, and three months later nobody can point to what actually changed — or the topic sits on the agenda for two quarters and never converts. Different failure, same cause.
The fix isn’t a better tool. It’s a structure with edges: three checkpoints, three deliverables, three honest go or no-go moments, instead of an initiative that can quietly run forever. Each phase answers exactly one question.
Days 1–30 · Diagnosis, not deployment
Where this actually applies, and who owns it. We audit what is already happening — shadow use on personal accounts, AI features already paid for inside a CRM or a writing tool, and the two or three workflows that cause the most recurring friction. The deliverable is one workflow named, one owner with real authority, and the number that will mean “this worked” at day 90. Nothing gets purchased.
Days 31–60 · One pilot, not a platform
Whether it works, in hours you can count. The one workflow runs for a full thirty days with protected time and a short feedback loop, not a rollout to three teams at once. We capture usage data and a real hours-saved or hours-reallocated figure against a baseline, and write down the friction rather than remembering it.
Days 61–90 · Scale, fix or stop
An explicit decision, taken on the day-60 numbers and written down — expand, fix, or stop cleanly. Plus the minimum governance a company your size actually needs: three short written answers on data, ownership of outputs and access. A real number reported upward with its baseline, and a named owner for what happens next. Then the same structure runs on the next workflow, faster.
The full method is published openly in the AI Adoption Sprint series, including the one-page tracker.
The sprint runs over 90 days alongside your existing team — it is not a full-time engagement. We set the working cadence in the first week, so every step survives limited time and no dedicated budget line. On site in Belgium or remote.
Format
90 days, three checkpoints
Investment
Pricing on request
No. The sequence works whether you land on an open-source, self-hosted stack or a commercial one, and I don’t sell either. That independence is the point — see the note on how the sprint treats the open-source question.
It is long enough to produce one decision with real evidence on one workflow. It isn’t a transformation programme, and it isn’t meant to be.
Yes, and it’s the most common starting point. The first month is diagnosis either way: name the workflow, the owner and the number, then run the pilot properly on what you already have.
The workshop trains a team on its own tasks over one to three days. The sprint is the ninety days around it: choosing one workflow, running it, and deciding what happens next.
Enough to be safe at your size: data handling, ownership of outputs and access, in three short written answers. Where a use case genuinely warrants legal or compliance advice, I’ll say so rather than improvise it.
During the sprint we use non-sensitive data only. For confidential data, I can set up models that run on your own infrastructure.
Tell me where AI sits today and what is taking the most time. We will pick the one workflow to start with.