Find Out Before You Launch
Most products that flop don't fail on the build — they fail because nobody wanted them, and the team found out months and a marketing budget too late. MiroFish is an open-source swarm-intelligence engine built to surface that signal early. Give it your launch page, your pricing or an ad, and it simulates how a crowd of thousands of AI customers reacts.
A Whole Virtual Market
MiroFish generates thousands of distinct AI agents — founders, engineers, students, designers, investors, parents and more — each with its own background, budget, memory and personality, then drops them into a simulated social world. Each one evaluates your material on its own terms, based on who it is, rather than one model giving you an averaged opinion. It's built on the OASIS engine from CAMEL-AI.
Feed It, Ask It, Read the Report
The workflow is three steps: upload your page, pricing or ad; ask one question, such as "love it or ignore it?"; and read the structured prediction report that comes back. It runs self-hosted on any OpenAI-compatible model, so your launch material stays on your own setup.
Treat It as a Signal, Not Gospel
MiroFish is an early-stage v0.1 prototype, and the guide is upfront about it: there are no proven accuracy benchmarks yet. Use it as a directional gut-check to form hypotheses quickly — then test the ones that survive with real users before you commit.
What's in the Guide
Who's in the simulated crowd, how the pipeline runs from your page to the prediction report, what self-hosting means for your data, the honest limits of a v0.1 tool, and a direct link to the repo.
The full field guide is in the PDF.
Get the Guide
Drop your email below and we'll send it straight to your inbox.