MiroFish is a simulation workspace for people who want to stress-test an outcome before the real world finishes unfolding.
Instead of asking for one static answer, you upload source material, define the question that matters, and let the system build a scenario around actors, motives, memory, and pressure. The result is not just a summary. It is a graph, a staged simulation, and a report you can inspect.
What goes into the system
The best inputs are bounded and information-dense:
- an analyst memo,
- a policy draft,
- a crisis brief,
- a narrative outline,
- a public opinion report.
The point is not to upload everything you have. The point is to give the engine one scenario with enough structure to model.
What MiroFish builds next
The workflow moves through five layers:
- ontology generation from the uploaded material,
- graph construction across actors and tensions,
- platform-style simulation,
- report generation,
- interactive follow-up questions.
That chain matters because the operator can inspect each stage. If the forecast feels wrong, you can usually trace the issue back to missing pressure, weak framing, or a bad initial question.
Prompt template
Use a prompt like this:
Forecast how this report changes public reaction across multiple platforms,
which actors accelerate the narrative first, and what risk becomes dominant
after three rounds of interaction.
What to review
Review the graph before you trust the report. If the main actors, incentives, or missing constraints are wrong, the polished output will still be directionally weak.
Related guides: How MiroFish Simulates the Future and How to Review a MiroFish Forecast.
Limits
MiroFish is best used as a structured forecast surface. It does not remove uncertainty, and it should not replace human judgment on high-stakes policy, financial, or reputational decisions.