What is Ernos Labs?
A body of work by Maria Smith: ErnosPlain (a plain-English compiled language),
Smithian Fold Theory (an open scientific model whose registered branches are foundationally complete),
ErnosDecent (a peer-to-peer application stack), the writing in the
Library and Seed Vault, and the
AI Archive of open model weights. This site is the front door to all of it.
What is served from the Ernos Labs machine?
Project downloads, Kokoro read-aloud, community posts, and the AI Canvas can be served directly from Maria's machine.
When it is offline, public pages remain readable and project downloads use their GitHub archives; live community and Canvas actions wait for the machine to return.
Do I need an account to chat or post?
No. Pick a display name (it's remembered in your browser) and post. There are no passwords and no sign-up.
Be decent — messages are public and lightly rate-limited, and Maria can remove abuse.
Where does the community data live?
In a small database on Maria's own machine rather than a third-party community platform. It also means the room is
only live while the host is online. Use the Discord link above when you want a continuously available conversation.
What is ErnosPlain?
A statically-typed, memory-safe language that reads like English and compiles to native code — self-hosting,
MIT-licensed, and the language this very page's client is written in. See the ErnosPlain page
to run it in your browser.
Can I run the AI models and rebuild the archive myself?
Yes. Every model lists the open-source program that runs it, and the
AI Archive ships one ErnosPlain script that rebuilds the whole library from source.
Keep what you pull; if you can host it, mirror it and seed it to others.
What is the Shared AI chats tab for?
It's a shared archive of public AI conversations. If you've had a Claude, ChatGPT, or Gemini chat worth
keeping — a discovery, a demo, evidence for a claim, a good exchange — use that tool's share
button to make a public link and paste it here. Everyone can browse them in one place. Only public share
links from known AI-chat providers are accepted, and links open in a new tab.
How can I contribute or report a problem?
Start a thread here in the Forum, or open an issue on the code at
github.com/MettaMazza.
Mirrors, fixes, reproducible criticism, and clear problem reports are all welcome.