The challenge: expert knowledge is not automatically learning knowledge

How can this be developed into a modern self-study course that works without a trainer? We worked on exactly this challenge in a specific project from technical documentation.

The starting point was 13 source documents on technical standards, EU regulation, product safety and CE conformity. They include a NormenGuide, a SnipBook and topic-specific material on risk assessment, user information, warnings and digitalisation.

The materials contained valuable specialist knowledge but were mostly designed for trainer-led courses. The new target group: employees without in-depth knowledge of European legislation and technical standardisation.

The aim was therefore not simply to convert the existing presentations into videos. The actual task was to transform complex expert knowledge into understandable learning units that build on one another didactically.

The solution: an AI-supported learning architecture

First, the specialist materials were mapped to the desired learning objectives. Then a learning structure was created that follows not the order of the source documents but the learners’ understanding.

The result of the design phase:

  • 11 learning modules with topics that build on one another
  • 360 minutes of planned learning time (six hours)
  • Short learning nuggets with clearly delimited topics
  • Different media formats depending on the learning objective
  • Defined source mapping for each module
  • A structured production prompt for Claude Code

The didactic thread is: CE → risk → documentation → warnings → application. Simplifying the subject matter must not lead to incorrect legal statements; the applicable requirements depend on the product and legal act concerned.

One piece of content – different learning formats

Not all content is conveyed in the same way. Several formats were therefore combined for the self-study course:

  • Explainer films with an AI avatar: an avatar explains the content using visually reduced presentation slides. Each slide has its own speaker text. The slides carry the core message; the avatar provides examples and context.
  • Podcasts: practical stories convey connections and show what consequences unclear or missing information can have.
  • Knowledge checks and quizzes: multiple-choice questions check understanding. The concept provides for questions with three answer options and no more than two correct answers.
  • Case studies, reflections and exercises: learners assess specific situations, spot mistakes and apply the content.
  • Structured learning slides: key statements, processes and comparisons are presented in visually reduced form and can later serve as a reference.

AI supports the transformation – experts ensure the quality

A key part of the project was developing a master prompt for Claude Code. It describes the target group, learning objectives, English as the output language, source documents, modules, media formats, time budgets and quality criteria.

The production logic is deliberately dynamic: not every module needs every format. Instead, only the elements planned for each module are to be produced in the intended order – from the avatar script to the quiz.

For visualisation, suitable existing illustrations from the sources are to be used in preference. Where they are missing, supplementary AI-generated images are planned, such as photorealistic work situations or reduced process graphics.

Important: for legal and normative content, statements, the status of standards, usage rights to illustrations and factual currency must be checked before publication. AI does not replace this approval.

What has actually been achieved?

The work so far has completed the design and production preparation, not the entire media production. The following were developed:

  • Source and topic mapping
  • A complete agenda with eleven modules
  • Time budgets and media formats for each learning nugget
  • Rules for slides, avatar texts, podcasts, cases, exercises and quizzes
  • A modular master prompt for further content creation
  • Initial image concepts for the visual implementation

Still outstanding are the full production and integration of all English-language learning content, the final expert acceptance and the measurement of learning success. For this reason, no gains in time, cost or performance are claimed.

The bigger benefit: using company knowledge several times

The approach is not limited to technical documentation. Companies often already own the content they need for employee qualification. The challenge is to prepare this information for the target group.

After suitable expert review, existing knowledge sources can give rise to various formats: self-study courses, explainer videos, podcasts, quizzes, reference information and knowledge modules for AI assistants.

The central idea: knowledge does not have to be rewritten from scratch for every format. It can be structured, transformed and reused.

What role does Yoomind play?

Yoomind’s approach is not just to store company knowledge but to make it usable for different applications. The Yoomind Workbench is geared towards creating different content formats; learning journeys structure this content into traceable learning processes.

The use case described shows how such a transformation can be designed and prepared – from existing specialist materials to a multimedia learning architecture.

Conclusion: the future of training begins with existing knowledge

Companies do not have to start from scratch to develop modern digital learning offers. The specialist knowledge they need often already exists. AI can help to structure it, present it clearly and transfer it into suitable learning formats.

Documents become structured knowledge. Knowledge becomes learning content. Learning content opens up new possibilities for employee qualification.