Local AI classroom
See how AI meets a life
A playground for retrieval, chains, graphs, and vector search. People stay in the foreground. Nothing you type is kept.
AI has felt like a black box for too long.
It is not about collecting more tools. It is about seeing the steps: a chunk, a neighbor, a citation, a hop on a chain, a node on a graph. This site is built so you can watch that happen on your own machine.
Five doors in
Start with a lab
Vector DB
Watch meaning become geometry. Query a local index and see the neighbors.
RAG
Retrieve first. Generate second. Read the passages the answer stands on.
LangGraph
A graph you can see: route, retrieve, draft, critique, answer.
LangChain
A pipe you can see: retrieve, fill a template, invoke, parse.
Agentic AI
A refill graph you can see: tools pick the path, a pharmacist click finishes escalate.
How it works
Learn. Try. See the trace.
- Learn
Each lab tells you what the idea is and why it exists, in plain language.
- Try
Ask a question of a bundled teaching corpus. Your words are not saved.
- See
Flow lights as each step finishes. AWS Architecture sits beside it as the production twin. Neighbors, citations, and node output stay next to the answer.
Top questions
You might be wondering...
Is it really free?
Yes. There is no paid API and no account. On your machine the labs use local Ollama. The GitHub Pages demo only replays canned teaching runs in the browser.
Does this call AWS?
No. Flow is the live lab on your machine. AWS Architecture is a static picture of how that same step could run in production. Click the title for the stacked twin. No AWS keys, no hosted models.
Does anything leave my machine?
Not in this version. Live labs talk to localhost only. The Pages demo never sends your words anywhere — it only replays recorded sample questions.
What happens to what I type?
It lives for the request, maybe a few minutes in memory — fifteen minutes if a refill is waiting for Approve or Deny — then it is gone. Restart the API and the slate is clean.