Claude training for companies in Berlin

Why Berlin teams train on Claude together

A team that pays for several AI tools does not yet share a way of working with them. One person writes long instructions, another pastes a ticket and hopes, and nobody checks a result the same way twice. The training gives the team one method for the tools it already has, and a Claude project keeps that method where everyone finds it.

What a Claude training for a Berlin team covers

The content follows your team’s work. Parts that fit a startup team:

• Projects: one project holds the product description and the current roadmap, so every chat starts with context that would otherwise be typed again and again.
• Instructions a colleague can reuse: the team writes down what it asks Claude, in which order, and how it checks the answer before the answer goes out.
• Skills for repeated work: a written procedure for release notes or support replies that Claude loads when the task matches.
• Connectors: Claude reads from the ticket system or the wiki through MCP servers, so nobody copies data by hand.
• For developers, in their own session: Claude as coding agent under rules that a hook enforces before each tool call.

Typical exercises for a startup team: release notes from a changelog, a support reply in English and German from one ticket, a product page from a feature list, and a specification drafted from a chat thread and then checked against the ticket.

Online from Frankfurt, or on site in Berlin

I work from Frankfurt. Online sessions run by video, which suits Berlin teams with colleagues in other countries and time zones. For an on-site day I come to your office in Berlin or Potsdam. Participants bring their own laptops and work with your company’s Claude accounts, so what they build in the session stays with them.

Both formats are built on your company’s tasks. The needs assessment decides which format fits and how the sessions are cut.

Tool inventory before the training

Your team probably does not start from zero. Before our call, it helps to know what is already there:

1. Which AI tools does the company pay for, and who uses which one?
2. Which tasks does each tool do well today, and where does the team switch tools in the middle of a task?
3. Where do drafts and instructions live: in a chat history, in a wiki, nowhere?
4. Which Claude plan does the company use, and who administers it?
5. Which data may staff enter? Ask your data protection officer.
6. In which language does the team write to customers, and in which does it talk internally?
7. Are there developers who need a session of their own?
8. Who keeps the team’s shared instructions current once the training is over?

Who I am

I am Michael Wutzke, an IT architect for data platforms in Frankfurt with more than twenty years in IT and media. At the Frankfurt-based company Blocksize Capital I was Head of Decentralized Finance and Node Operations and then Chief Information Officer. I have more than ten years of leadership experience, with teams of up to 50 employees. I teach at Claude Hacker House, 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. I work in German and English, and I am interested in open-source AI models that a company runs on its own servers. Details: Teaching and certifications.