AI training for the staff of companies in Germany

Why companies train the whole staff

Few companies train all of their staff. In a Bitkom survey of 603 companies with 20 or more employees, published on September 14, 2026, 11 percent trained all their employees in the use of AI, 22 percent most of them and 37 percent selected employees. A quarter offered no training at all (Bitkom).

A training for selected staff leaves the others to find their own way with the tools, and nobody tells them which data stays out. A program for the whole staff gives each person the same basics and the same rules, and then time on the tasks of their own department.

A training plan by department

The basics are the same for everyone. After that each department practices on its own work. The plan shows typical exercises; the needs assessment replaces them with the tasks your departments name.

Group	Exercise from its own work	Rule it practices
Everyone	Write an instruction that gives the same result twice, and check an answer for names and numbers	Which data stays out of the tool
Management	Condense a long report into one page for a decision	The person who signs keeps the responsibility
Sales	Draft an offer text from the notes of a customer call	Prices and terms come from your own documents
Accounting and finance	Explain a deviation in a monthly report in plain words	Each figure gets checked against its source
HR	Draft a job ad or an onboarding document	Decisions about applicants and employees stay with people
Customer service	Draft a reply from your own answer templates	A person sends each reply
Marketing	Write in your company’s style, from a style guide stored in a project	Product claims get checked before publication
Developers	Claude or Codex as coding agent, in separate sessions	Rules and checks in the repository

For developers the rules around a coding agent are a subject of their own: Agent harness engineering.

How the program runs online

An online program reaches staff at several sites with one set of material. It has these parts:

• A basics session for everyone, in groups small enough that each participant works in the tool.
• A session for each department, with the exercises from the plan above.
• A question hour some weeks later, for what came up in daily work.
• A contact person in each department who collects questions in between.

On-site days suit teams that want to practice in the room, in Frankfurt Rhine-Main or at your office elsewhere in Germany.

Who I am

I am Michael Wutzke, an AI engineer and trainer in Frankfurt with more than twenty years in IT and media. At Claude Hacker House I teach practical work with Claude together with Fayssal El Mofatiche and Christian Schulze; my sessions cover prompting, AI-assisted workflows, MCP servers and virtual organizations. In client work I build software with Claude and Codex, and I am interested in open-source AI models that a company runs on its own servers. I hold a training authorization from IHK Frankfurt for two IT apprenticeships, and I take part in Claude Certified Architect Foundations, Anthropic’s first certification program. Details: Teaching and certifications.