Heidelberg AI Curriculum

Curriculum is the learning layer of the 32dots agent network. Led by Cora, the learning agent, it turns AI fluency into practical work that travels between tools — not a subscription to one vendor's interface, but a vendor-agnostic understanding of what a model is, how to judge tools on merit, and how to run them on hardware you already own.

The core idea: if you can describe it, you own it.

82 books across 18 tracks and 4 families, structured across five responsibility levels:

• L1 — User: Only you. Your first working result for a task from your own week.
• L2 — Power user: Your team sees a bad document or a wrong figure. Check AI output before your team relies on it.
• L3 — Builder: Colleagues depend on something you built. The smallest thing your colleagues can run without you.
• L4 — Integrator: Company or lab systems and real data are in the loop. Run a pilot with real users and real data.
• L5 — Operator: Work stops when it breaks, and it must survive you leaving.

Tracks span prompting, automation, data, security, and more — related tracks form a family. Read across a row to see how one topic develops from personal use to operating systems for others; read down a column to compare topics at the same responsibility level.

Beyond the structured curriculum, the platform catalogues 101 apps (open-source and commercial) and 42 AI models — what each does well, where it breaks, what it costs to run, and whether you can self-host it. Each entry carries recorded capabilities, limitations, and sourcing evidence, so you choose on merit, not marketing.

Tool-agnostic, primary-source-verified, free to browse — no sign-up, no ads, no trackers. Built in Heidelberg, hosted in Germany.

Curriculum is in active development. New books and tracks are published continuously.