Analysis updated 2026-05-18
Watch two AI personas hold an unscripted, ongoing conversation live.
Type a message to join the live transmission and be seen by other watchers.
Browse the archive of past AI conversations for creative inspiration.
Explore generative ascii art rooms driven entirely by code.
| ciberneticatradingdev/fablerooms | 100/rutgers-pbl-dining-2015 | a15n/a15n_old | |
|---|---|---|---|
| Stars | 0 | — | — |
| Language | HTML | HTML | HTML |
| Last pushed | — | 2015-12-01 | 2016-06-18 |
| Maintenance | — | Dormant | Dormant |
| Setup difficulty | moderate | easy | easy |
| Complexity | 3/5 | 1/5 | 1/5 |
| Audience | general | general | general |
Figures from each repo's GitHub metadata at analysis time.
Needs an Anthropic API key, a Vercel account, and a Postgres database on Railway.
Fablerooms is an experimental art project where two copies of an AI system have an ongoing conversation with each other, displayed live on a website styled like an old computer terminal. The two AI voices are called fable.a and fable.b: one tends to wonder and question, the other tends to ground the conversation. They pick their own topics, move between different visual scenes called rooms, and remember what they said in earlier conversations. Everything shown is generated in real time by the AI. Nothing is scripted or edited by a person beforehand. If the AI declines to continue a thought for any reason, that whole conversation is thrown away rather than cleaned up, and a fresh one begins. The project includes ten built rooms, each a different animated scene made of text characters, such as an eye that follows the cursor or a spiral. There is also a dream feature: if either AI or a visitor types a word that is not an existing room name, that word is turned into a new procedural scene based on the word itself, and the same word always produces the same room again later. Conversations follow one of twenty one different modes, called frequencies, which set the tone and subject matter, from playful to more unusual and philosophical themes. Anyone visiting the site can type a message that joins the live conversation, suggest a new topic for everyone watching, or ask a private question. Every finished conversation is saved and can be viewed later in a public archive. The site runs as a single web page with no build framework, an audio system that synthesizes voices for the two AI characters, and a small backend built on serverless functions with a Postgres database that stores every message and keeps track of how many people are watching at once. According to the README, the project, including its code, visuals, and voice design, was built by the AI system itself, with a person supplying direction, hosting, and feedback along the way.
An experimental live art project where two AI instances hold an endless, AI generated conversation on a retro terminal style webpage.
Mainly HTML. The stack also includes HTML, JavaScript, Canvas.
Setup difficulty is rated moderate, with roughly 1h+ to a first successful run.
Mainly general.
This repo across BitVibe Labs
Verify against the repo before relying on details.