In eight weeks, you build the system that reviews your technical documentation.

No. If you can work with Word and copy text, that is enough. To make sure, our Head of HR, Susanne Exler, completed the entire programme in advance — and succeeded.

No — and that is intentional. In the eight weeks, you build your system for two review aspects: intended purpose and risk management. These two are the worked example on which you learn the method — from the review criterion through the data structure to the validated run.

For each further review aspect, you follow the same path again. The method is identical, and you can apply it on your own after the programme; what you will then be missing is not the skill but someone sitting next to you in the workshop. If you want to save yourself the work, you can request 13 additional aspects as fully worked-out material via our sales team (Extended IP Package) — but you don’t need to.

No — and that is a deliberate decision. Your review system should work in a traceable and repeatable way: fixed review criteria, fixed data structures, a documented process. That is exactly what can be validated. A freely acting agent cannot be validated. We use agentic tools where they belong: when building the system. Reviews are then carried out according to a fixed procedure for which you are responsible — you wrote the yardstick, you decide — and no AI takes that responsibility off your hands.

That is the central challenge, and it took up most of the past nine months. It is not brought under control by a better model but by craft: fixed review criteria, fixed data structures and a measurement that shows you how stable your system is. You build exactly this measurement in week 7. Reproducibility is not a by-product here — it is the topic.

Each week, a 45-minute live session and 1–2 hours for your assignment. In the Power User and Transformer Team Packages, you start the assignment in the workshop — a fixed 45-minute slot in the joint group, with me at your side. In the Explorer Package, you do this part on your own. If you miss a week, the model solution gets you back on track.

All live sessions are recorded and available in the course area. The workshop is not recorded; with the weekly model solution you can still pick up where the others are.

Access is included in the programme — IP-protected, without an IT project. Copilot is made for writing, not for systematically reviewing entire files — that is what you build your own system for here.

In the course, you do not process any data from your company — all work is done exclusively on the example. For the AI working environment in general: inputs are processed by the provider (Anthropic Ireland, Limited), including outside the EU. There is no training on your data, and the rights to the results belong to your company. We do not guarantee processing exclusively within the EU.

The details are in the approval package for your IT — you can download it above and forward it directly. It backs up the commitments with references and also states what is expressly not guaranteed.

No. In the course, everyone works exclusively on the example technical documentation of a medical device — complete, with all files. Your company’s documents do not belong in the course environment; this is governed by the special terms and conditions. You transfer the method to your own files after the programme — using your company’s own AI contract.

Because otherwise you cannot evaluate what you buy — and because every tool has to be extended with your own review criteria. You learn both here.

Yes, the Transformer Team Package covers 3–5 people and costs a flat €20,000 plus VAT, regardless of how many places you fill. You book it via our sales team — who can also provide an offer for larger groups.

“Catch the Train” is a learning programme: you build knowledge you can apply to all tasks. Here you build something — it is exclusively about TD review, and at the end you have a working system on your computer. In addition, the three hurdles where participants typically get stuck have been removed: no time (fixed dates, weekly model solution), no AI access (it’s included), no concrete project (it is fixed from day one).

There is no sit-down exam. You earn the “Certified MedTech AI Practitioner” certificate through participation, a question set (multiple choice) in the course area and a passed proof of competence — which is what you build in week 7 anyway: your validation file, plus a small benchmark of your own in which you slip deliberately planted defects into your system and measure whether it finds them. What is assessed is not the detection rate but whether you can demonstrate how well your system works. You can submit up to eight weeks after the end of the programme; certificates are issued in February. If you don’t pass, you can retake it once free of charge within three months.

Talk to me after the first week — we will find a solution together. In the previous programmes there were neither complaints nor withdrawals. Cancellation is free of charge up to four weeks before the programme starts; everything else is governed by the special terms and conditions. If a registered person leaves your company, you can nominate a colleague once per seat.

Yes. The special terms and conditions and the data protection information for participants apply — you can download both above. The key rules: work only on the example, do not enter personal data, do not share your personal access. Before the course starts, each participant confirms these usage notes.

No. Zero to Hero is sold out: all 50 seats have been taken and the sales window is closed. If you are interested, please leave us your contact details via the contact form.