Source code, notes, transcripts, data and reading go in. Working lessons, runnable practice and interactive worlds come out, every claim checked against the material it came from. Worlds is live. The rest of the Brain is arriving.
The core does the hard part: it reads material, proves what it found, and refuses to assert anything it cannot trace. Each integration is a different thing to point that at. Click any of them.
A course becomes a place learners work. Not a video and a notebook: a browser environment running the instructor's real code on the real data, with exercises that cannot ship unless they run and quotes that cannot ship unless they match the recording.
The part that matters. A theme cannot contain a number, so every figure a learner sees resolves from the data. An exercise cannot publish unless its reference solution passes and its untouched starter fails. A quote cannot publish unless it matches the recording character for character. Where a recording and its code disagree, the Brain raises it and a person decides. Generated material that nobody approved never goes live.
Every integration below reuses what Worlds proved: read the material, verify what you found, hand it to a person. What changes is the material and what comes out the other side.
Twenty minutes. Bring a repo, a lockfile, or a recording of something you teach. We would rather run it than describe it.