AI expert for your team in Germany

What an AI expert does in your company

An AI expert is the colleague the others ask before they hand work to AI. In a company without one, staff try ChatGPT on their own, and nobody decides which data goes into it. The expert makes that decision with management and sets up the tools that follow from it.

KfW Research reported on February 11, 2026 that 20 percent of Mittelstand firms used AI in the period from 2022 to 2024. Firms with their own research and development used it at 53 percent (KfW Research). A firm without a research department can bring that knowledge into the team through one hire.

Questions to ask an AI expert before you hire, with my answers

A company that hires its first AI expert often has nobody in house who can test a candidate. These questions help, and my answers come from client work and teaching. Bring your own questions to our call.

Question	My answer
Which AI tools do you use in your own work?	Claude daily, Codex and the Gemini CLI for code. The three read their rules from one shared file, so a rule is written once.
What have you built with AI that runs in production?	For a client, several agent sessions and the human team work in one repository at the same time and break up a monolithic platform.
How do you stop an agent from changing live data?	At the resource: its database role only reads, its tokens work on the development system, and the production branch is protected. A person confirms a production change outside the chat.
How do you know that an agent’s report is right?	A read-only script measures the result. In an early job an agent did the work right and reported a count for a wider area than the task named. Since then the script’s number counts.
Which decisions stay with a person?	Changes to live data, questions of classification and each new type in the data model.
How do you teach staff?	In workshops where participants work with the tools in the room, as at Claude Hacker House. I hold a training authorization from IHK Frankfurt for two IT apprenticeships, application development and system integration.
What have you not solved yet?	Which decisions can leave the human queue without a loss of quality, and how to measure that. It is the main open question of my research on virtual organizations.

The rules behind these answers are on AI agents under the same rules as humans.

Profile at a glance

Role	AI expert, AI specialist, AI enablement lead
Engagement	Permanent, remote for the most part, on-site days in Frankfurt Rhine-Main or at your office; freelance; interim
Location	Frankfurt, Germany
AI tools	Claude, ChatGPT, Codex, Gemini CLI; Paperclip for agents with roles; MCP servers
Teaching	Lecturer at Claude Hacker House; training authorization from IHK Frankfurt for two IT apprenticeships
Languages	German, English

Who I am

I am Michael Wutzke, and I have worked in IT and media for more than twenty years. I was Head of Decentralized Finance and Node Operations, then Chief Information Officer, at the Frankfurt-based company Blocksize Capital, and I held earlier leadership roles as IT Team Lead and Managing Director. I teach at Claude Hacker House, where my sessions cover MCP servers and virtual organizations of agents, and I take part in Claude Certified Architect Foundations, the first official certification program of Anthropic. Besides Claude I work with Codex, and I am interested in open-source AI models that a company runs on its own servers. Details: Career stages.