MiroFish is an AI prediction workflow product whose core idea is to "ask questions like using ChatGPT, but what runs behind the scenes is a simulation." Users can ask a question directly and let the system complete seed generation, simulation runs, and report output as one continuous prediction workflow in the background, while preserving a single-conversation experience. The product can turn text, PDF, MD, and TXT files into graph building, simulation, reports, and subsequent chat.
In terms of main capabilities, MiroFish starts from users' natural-language questions, reports, policy drafts, market notes, or story fragments, extracts actors, relationships, pressures, and factual anchors, and lets agents reason based on structure; it then lets roles interact over multiple rounds on short-form content and thread-style social surfaces; it then condenses emergent behavior into turning points, risks, confidence signals, and follow-up paths; finally, it allows users to continue asking questions about the generated world instead of stopping at a static answer. Below each answer, a structured result card is provided, including a summary, report entry point, and follow-up paths.
In terms of usage, users can start with a clear question and decide whether supporting files need to be added, without losing the speed of the conversation. The product recommends clarifying the decision, audience, possible triggers, and time horizon when asking questions; narrower questions make the simulated world drift less. It works best when PDFs, Markdown, and text files contain specific actors, incentives, constraints, or prior context. Users can treat the output as decision support and focus on resistance signals, narrative bridges, and assumptions worth verifying with real data.
Applicable scenarios include: simulating how audience groups will amplify, resist, or reinterpret campaign messages before committing budget; modeling different customer segments' sentiment, value perception, and possible paths of opposition before a product change is announced; using simulation as a tabletop exercise for controversy, coalition formation, and second-order reactions; stress-testing market stories where spreadsheets easily overlook feedback loops among analysts, retail attention, and public discourse. Specific use cases cover launches, pricing changes, policy debates, market narratives, crisis response, and creative continuation, among other scenarios involving human reaction loops.
In terms of limitations, MiroFish is positioned as exploratory decision support, not a precise prediction tool, and users should use it together with judgment, analysis, and real-world validation. The product emphasizes its exploratory positioning and does not claim accuracy. Users can start from text alone without providing files, and later add files to ground the simulation in specific reports, briefings, or source documents. Information about fees, pricing, and specific platform support is not currently provided in the materials.
