Services
Assess & scan
Together we map how your knowledge work runs: where the thinking sits, where the routine work sits, and what that routine work costs in hours and euros. AI is strong at some steps and weak at others, so we draw that line step by step.

The Working Session
Ninety minutes along your own workflow. Three observations and one recommendation, on paper.
€ 399
Ninety minutes with the leadership or the core team, on site or online, and no longer. We walk through your own workflow chronologically: from brief to delivery, where AI already touches the work, where the team sighs, and where nobody dares.
After the walkthrough we look at both sides live, on your own examples: what can safely go faster today (collecting, sorting, preparing) and where credibility is at stake, what this actually rests on. Equal weight to both. In the last fifteen minutes we name three observations and one recommendation out loud.
Most conversations about AI stay stuck in what might be possible. This session is about what happens at your organisation, in the order it happens. By the end you know where the time leaks away and where the trail runs thin, and you have it on paper, whether we continue together or not.
The steps
- Beforehand: you send a short note on the kind of work and who will attend.
- The walkthrough, 45 minutes: your own workflow from brief to delivery.
- The two sides, 30 minutes: where the time sits, and where credibility is at stake.
- The first picture, 15 minutes: three observations and one recommendation, out loud and together.
- Within two working days: one A4 with that same first picture.
What isn't included
A measurement. These are observations from ninety minutes, not an audit. No benchmark, no cost estimate, no report. That is the Accountability Scan, and the € 399 is credited against it if you continue within three months.
The Sorting Session
One half-day to sort all your material into what can and can't go through AI.
A guided half-day working session in which we sort all of the organization's material into two bins: documents with personal data, and knowledge without it. Plus the three measures that remove the biggest risk that same week.
After the session you have a sorting map of your own material: what can't go into an AI service for now, and where the safe gains are. The three no-cost measures are set up right away: organization accounts instead of personal ones, two-factor authentication, and a one-page agreement on what may never go into an AI service, who signs off, and where to report a mistake.
Most organizations ask the wrong first question: which tool is safe? The useful question is which documents can be processed with AI at all. Skip the sorting and you have to apply the strictest rules to everything, and then, in practice, nothing is allowed. Sort it, and you can safely start tomorrow with the part that has no names in it. Half a day of work, and the whole further approach is settled.
The steps
- Beforehand: a short questionnaire about what material exists and where it's stored.
- The session (a half day, on-site or online): we go through the material together, one question per type. Does it contain a name or an identifiable person? Doubt means bin one.
- Right after: setting up the three measures, together with whoever manages the accounts.
- Within a week: the sorting map and the agreements sheet, ready to use.
What isn't included
A legal judgment on the lawful basis for processing; that remains work for the data protection officer or lawyer. The map does make that conversation concrete, though.
The Second Opinion
An independent assessment of the proposal, contract, or system on the table.
An independent assessment of one AI proposal, contract, or system on the table. Against the six layers, including the contract side: the data processing agreement, data storage, and whether the vendor trains on your input.
A report with a verdict per layer, the concrete risk of every weak spot, and the questions the supplier still needs to answer. It closes with clear advice: proceed, adjust, or reject, and why.
AI vendors sell well, and there's no one on your side of the table who can judge the build. That's exactly the gap expensive mistakes slip through: a chatbot with no boundaries, a system that invents its sources, a contract that stays silent on the data. One day of investigation is cheaper than a year stuck with a system that doesn't hold up. And sometimes the outcome is reassuring: this proposal checks out, go ahead and sign.
The steps
- You send the proposal, the contract, or access to the system.
- Assessment against the six layers of TRACES, with evidence per layer; where needed, we put the questions to the supplier ourselves.
- Within five business days: the report with the recommendation.
- A half-hour conversation to walk through the report, so you can carry on independently.
What isn't included
Negotiating with the vendor on your behalf, and legal contract drafting. The report does give your lawyer exactly the points that need adjusting.
The AI Accountability Scan
The full audit of your AI use and knowledge work, with an order in which to tackle it.
The full audit of the organization's AI use and knowledge work. Four to five days of investigation, closed with a report in plain language.
The Supplier Scan
Name a tool and we find out what the supplier publishes about it, with the source alongside.
Free supplier scanA fit for you?
In half an hour we'll see which of these fits you, and whether you need one at all. No sales pitch, an honest picture.
Book an intro callHere's what that looks like in practice.
Three clients, one rule: the AI does the work, the human approves.
Read the case