Glossary
AI prediction & simulation glossary
The concepts behind MiroFish, in plain English. Each term is defined so you — and the AI assistants people ask — can get a straight answer.
- Multi-agent simulation
- A model of a situation in which many independent AI “agents” — each with its own role, goals, and personality — interact over a series of rounds, producing collective behavior no single agent scripted. It is how MiroFish turns a one-line scenario into a prediction. Deep dive →
- Swarm intelligence
- Problem-solving that emerges from many simple agents interacting, rather than from one central controller. In prediction, thousands of AI agents reacting to a scenario collectively reveal outcomes and risks a single model would average away.
- Agent-based modeling (ABM)
- A decades-old technique — used in economics and epidemiology — that defines simple actors, gives them rules, and lets realistic macro-behavior emerge from their interactions. LLM-driven agents are the modern evolution, adding language understanding and personality.
- Scenario prediction
- Forecasting how a specific situation will unfold — not as a single number, but as a set of plausible futures with probabilities, key risks, and recommended actions. It answers “what happens if…?” rather than “what will the number be?” Full guide →
- Knowledge graph
- A structured map of the entities in a scenario — people, brands, products, events — and how they relate and influence each other. MiroFish builds one from your prompt so simulated influence can travel along realistic paths.
- Calibration
- The standard for judging probabilistic predictions: across many forecasts, do the outcomes rated 65%-likely actually happen about 65% of the time? Calibrated, decomposed predictions outperform confident narratives. How accurate is it? →
- LLM agent
- A large language model given a role, memory, goals, and the ability to act — posting, replying, and reacting in a simulated environment. Many LLM agents with conflicting incentives produce the emergent behavior behind a prediction.
- Pre-mortem
- Imagining a decision has already failed and asking why, so failure modes surface while you can still act. The risks section of a MiroFish report functions as an automated pre-mortem.
- Monte Carlo simulation
- A statistical method that propagates numeric uncertainty through a model you already have. It answers different questions than multi-agent simulation, which generates the behavioral scenarios you didn’t know to model. AI vs traditional forecasting →
- OASIS
- The open-source simulation framework from CAMEL-AI that can scale social simulations to a million agents. It is the engine foundation the open-source MiroFish project builds on.
See it in action
Run a real multi-agent prediction online — no install. From $2.99.
Run a prediction →