Getting AI to Actually Read Your Docs (So It Can Help You Build)

​Most of us have asked an AI assistant a question about a tool, only to get a confident, plausible-sounding wrong answer. Jürgen Haas has spent the last several months tackling that problem head-on for ECA — building out an MCP server, an agent skill, and a full documentation overhaul so that when you ask AI for help building an ECA model, it actually knows what it's talking about.

​Jürgen will walk us through:

• ​What MCP actually is and how it lets AI agents query live, structured documentation instead of guessing from stale training data

• ​A live show-and-tell using the ECA Guide as the example — what it looks like when an AI assistant can genuinely help you build a workflow model from natural language

• ​Other "robot-findable" docs strategies (llms.txt, agent skills, structured markdown, etc.) and an honest discussion of what's actually paying off versus what's still unproven

​If you maintain documentation for a module, product, or team wiki and you're wondering whether it's time to make it AI-legible — this is the session for you.