Quickstart¶
Welcome. This is a slow, friendly, hands-on tour. We'll build a handful of agents, each a little more capable than the last, and explain everything along the way, including the AI models themselves. By the end you'll have run real agents on a real peer-to-peer network and understood what every line does.
You do not need to be an AI expert
Don't know what a language model, a ResNet, or a tensor is? Perfect, that's who this is written for. Whenever a piece of jargon shows up, there's a little "What's a …?" box you can expand. Experts can skip those.
How this quickstart works¶
It's split into four episodes of rising difficulty. Each one runs on its own and ends with a recap, so you can stop whenever you've had enough.
| Episode | You'll build | New ideas |
|---|---|---|
| 1 · A chatbot in one file | A small AI you can chat with | agent · node · lone wolf |
| 2 · Talk to it from Python | A second agent that reaches the first | a second node · connecting · interacting |
| 3 · Two agents, fully automatic | Agents that exchange data with no human | typed streams · automation · a vision model |
| 4 · Where your story goes next | (a map, not code) | worlds · learning · your own model |
Before you start
You need the SDK installed and a free token. If you haven't yet:
pip install unaiverse, see Installation.- Get a token at unaiverse.io, you'll be asked for it the first time you run an agent (then it's remembered).
That's it. Everything below is copy-paste.