What It Is
Before you publish something, wouldn't you want to know how the world reacts? 4ort.win lets you find out in advance: feed it any document and hundreds of AI agent personas simulate public reaction, hour-by-hour, with opinions forming, spreading, and shifting over simulated time rather than a single static verdict.
Under the hood it's a lab fork of MiroFish, rebuilt to fit lab infrastructure: personas live in graph-native storage on Neo4j instead of flat records, sign-in runs through lab SSO, and the simulation runs on local LLMs via Ollama, so the whole crowd is self-hosted.
The killer pairing is with 4ort.life: substitute your actual network (the people from your private knowledge graph) into the simulation, and the question shifts from "how would a generic public react?" to "how would my people react?"
Highlights
- Document in, reaction out: feed it anything and watch simulated public opinion unfold
- Hundreds of personas: AI agent personas react hour-by-hour, not as a single static score
- Graph-native persona storage: personas live in Neo4j as a graph, not flat records
- Lab fork of MiroFish: open-source foundation, rebuilt for lab infrastructure
- Self-hosted simulation: the crowd runs on local LLMs via Ollama
- Pairs with 4ort.life: substitute your actual network into the simulation
Status
● Live (gated).