The workshop works with common tools like Cursor, Claude Code, or GitHub Copilot in agent mode. If you're already paying for AI coding tools, chances are they support agentic features your team isn't using yet. We'll discuss your current setup in the intro call.

No. Teams of 3-5 developers often see the clearest impact because there's less coordination overhead. The approach scales, but small teams can adopt it faster.

No. AI agents are good at writing boilerplate, tests, and repetitive code. They're bad at understanding business context, making architectural decisions, and knowing what to build. Your developers become more productive — they don't become unnecessary.

AI coding agents work with virtually any programming language and framework. Legacy code is actually a strong use case — agents can help with writing tests for untested code, understanding undocumented systems, and safe refactoring. We'll assess your specific situation in the intro call.

Valid concern, and one we address directly in the workshop. We cover which tools send code to external servers, which can run locally, and how to configure them for your compliance requirements. You'll leave with a clear picture of what goes where.

That's the whole point of working with your own code instead of demo projects. Your team builds configurations and habits they can use the next morning. You also get a written playbook with everything we set up, so nothing depends on memory. If your team gets stuck later, we can do a follow-up session.

I've been working exclusively with AI coding agents in real client projects for over a year. I also built an entire product — einfache-erechnung.de — through agentic coding in a few weeks. And I presented this approach at JavaLand 2026, one of the largest Java conferences in Germany. This isn't theory for me. It's how I work every day.