An open course practises on invented examples, inhouse AI training practises on your real processes. That is exactly why more sticks with us. Your team does not build on a demo, but on a problem you actually have, with your data and in your infrastructure. We plan scope, challenges, roles, data, tool access and the run of show so the training can be ready to start within one week, on-site or remote. What ships and when depends on the agreed scope.
Audience: companies that want to train on their own data and value data sovereignty, especially in regulated areas.
Inhouse AI training happens on your real processes and data, on-site or remote, and stays in your infrastructure. In our format it runs as a facilitated hackathon that produces deployable prototypes and a documented AI literacy measure, instead of a standard course with invented examples.
The problem with an open course
An open AI course can only work with general examples, because ten different companies sit in the room. Four things get lost: the link to your real, often sensitive case; your real data staying outside where the value actually forms; the data question staying unsolved until after the course; and the result never travelling with you as knowledge without a tool that runs in your context. Inhouse AI training flips that, because it starts on your real processes. What runs at the end fits your reality and therefore keeps getting used.
What the inhouse training includes
We build the training around your tasks. Beforehand we jointly pick the challenges that cost you the most time or money, and in the hackathon we work with your real data, in your rooms or remote over shared working environments. The stack of Cursor, Lovable, n8n, Gamma, Figma Make, Claude Code and Custom GPTs runs on your systems and access, across four stations: scoping on your processes, tool workshop, build sprint in your environment, and pitches with a handoff to a named owner.
Data protection and data sovereignty
Because it is about your real data, everything stays in your infrastructure. We build on your systems and access, nothing leaves unasked. That fits the GDPR and leaves you your data sovereignty. Especially for regulated areas, that is the reason inhouse AI training beats the open course. We settle the ground rules beforehand with your IT and privacy so no data protection issue slows the sprint.
Learning goals and benefits inhouse
3 to 5 deployable prototypes that fit your context instead of knowledge without application; trained AI champions in your own house who carry the prototypes forward; full data sovereignty because everything stays in your infrastructure; and a documented AI literacy measure toward EU AI Act Article 4, evidenced by the prototypes built inhouse.
Who it fits, and who it does not
| Good fit when | Not a fit when |
|---|---|
| you want to practise on your real processes and data | general examples in an open course are enough |
| data sovereignty matters and everything should stay in your infrastructure | you cannot provide your own systems or access |
| there are champions in-house to carry the prototypes forward | no one takes responsibility after the sprint |
| you want to work on-site or remote in your environment | you deliberately want a vendor-neutral open course |
Deliverables
Pre-scoping of the challenges on your real processes and data, a curated tool stack per challenge running on your systems and access, a facilitated tool workshop and build sprint on-site or remote with support, 3 to 5 deployable prototypes that fit your context, an impact report, skills library and IT handoff package per prototype, trained AI champions in your own house, and documentation of the AI literacy measure for participants.
Formats
| Format | Duration | Participants | For |
|---|---|---|---|
| Spark | 1 day inhouse | up to 10 | one deployable prototype on your data |
| Ignite | 2 days inhouse | 10 to 15 | several challenges in parallel, IT handoff package |
| Blaze | 3 to 5 days inhouse | 15+ | multiple teams, production-ready prototypes on your data |
Pricing on request for now.
Open course vs. inhouse AI training as a hackathon
| Criterion | Open AI course | Inhouse AI training as a hackathon |
|---|---|---|
| Practice material | invented, general examples | your real processes and data |
| Data sovereignty | data stays outside | everything stays in your infrastructure |
| Output | knowledge without a tool | deployable prototypes in your own context |
| Transfer to daily work | hard, out of context | direct, built on your real case |
| Champions in-house | rare | trained and named |
| Time to first result | depends on the course plan | ready in 1 week, outcome depends on scope |
Inhouse amplifies the effect, because hands-on here means your real data and tools. Full comparison on AI training vs. AI hackathon.
How the inhouse training can add up (a model, not a client number)
Use your own figures, on your real processes. Three inputs and one formula: weekly time for a manual step tied to your data, the share a prototype built on your data realistically removes, and internal hourly rate times number of people running the same process. Formula: hours saved per week times hourly rate times 45 working weeks times people, minus the one-off cost of the training and rebuild weeks. If document review costs six hours per person per week and a prototype removes half, that is three hours; across twelve people and a year that is an order of magnitude you can test the investment against honestly. Interactive calculator: /en/ki-schulung/tools/roi-calculator/.
Social proof
Teams from companies such as YOYABA, Onventis and NavVis have worked with us. Names, logos, workshop photos and public feedback from these teams are in the case studies.
FAQ
What is the advantage of inhouse AI training over an open course? The link to your daily work. Instead of invented examples, your team builds on your real processes and data. The result fits your reality, keeps getting used and delivers provable value instead of a certificate.
Does our data stay secure? Yes. Everything stays in your infrastructure, we build on your systems and access. That fits the GDPR and leaves you your data sovereignty. We settle the ground rules with your IT and privacy before the sprint so no data protection issue slows things down.
Does the inhouse training run on-site or remote? Both are possible. On-site we work in your rooms, remote over video and shared working environments. What fits better depends on team and location and is settled in the intro call.
How many employees can take part inhouse? Spark covers up to 10, Ignite 10 to 15, Blaze 15 and up. With larger staff, several teams work in parallel on their own challenges on your data.
What do we need for preparation? Access to the relevant systems and the challenges that cost you the most. We prepare environments and access on your systems so the sprint starts right away. From first contact to a finished prototype takes one week.
Do we get a record? Yes, documentation of the practical AI literacy measure, evidenced by the prototypes built inhouse. That documents a measure with a format that sticks on your real data.
Legal note: A hackathon and the competence record described here can document practical AI literacy measures, but they are not an officially mandated certificate and do not automatically guarantee compliance with Article 4. The company must assess its overall program for the relevant roles, context and risks.
