Learning

Learning turns your own material into courses. A course factory drafts the course, the learner works through it in a LearnHouse player with exercises, and a tutor answers questions along the way. It runs on our own server and is reached through its own interface, not through the Brain.

How it is built

LearnHouse

The learning platform the learner sees: courses, the player and accounts. We run our own fork of it, alongside a fork of PenEcho.

Learning services

Our layer on top: an API, an MCP server for agents, a background worker and an embedding service, with Postgres, Redis and a vector store behind them.

Course factory and tutor

The model work — drafting a course, tutoring, grading — goes to agents in the OpenFang runtime on a workstation. Embeddings are computed there too.

What works today

• ·Courses with practice exercises and a tutor, in the LearnHouse player
• ·Drawing exercises inside the player
• ·Video lessons produced with the Video space
• ·Customers as user groups inside an invitation-only organisation

Licence

The space builds on two forks, LearnHouse and PenEcho, both licensed AGPL-3.0. Our learning layer will be released under AGPL-3.0 as well; the licence file has not landed yet, so until it does the repository carries no licence. See Licensing.