Hjemmelegene

Hjemmelegene wanted to explore what AI tools could specifically mean for their development team. Not as a replacement for professional expertise, but as a tool to move faster from idea to functional code – and spend less time on routine tasks. Miles conducted a course in agentic coding with Claude Code, tailored for developers who want to adopt these types of tools in a professionally sound and structured way.

‍Practical introduction with a hands-on focus

‍The course was structured with short theoretical sessions alternating with practical exercises, allowing participants to try things out as they went. Topics ranged from a basic introduction to what agentic coding is and how it differs from standard use of language models, to more advanced subjects like using CLAUDE.md for project context, MCP servers for integration with external tools, and automation via Agent Skills and sub-agents. A central question for many participants was control: what actually happens when an AI agent suggests changes to the codebase, and how do you maintain oversight? The course addressed this directly by reviewing agent modes, context management, and commands that give the developer real control over the process.

‍For both beginners and experienced developers

‍The course was suitable for participants at various levels. For those new to Claude Code, it provided a solid foundation. For those with experience from similar tools like Cursor, the value often lay in seeing new work patterns and discovering possibilities they hadn't explored in their own projects.

‍The result‍

The participants from Hjemmelegene left the course with a concrete foundation for using agentic coding as a natural part of their development work – not as an experiment on the side, but as a tool integrated into their regular workflow. From exploration and refactoring to documentation, testing, and automation, with the developer still in the driver's seat.