Reference
Simulation
Scanna's proprietary multi-agent simulation engine for scenario modeling (private beta).
Private beta
The simulation engine is built and running internally — it's in private beta and not yet publicly available, so there's no public endpoint to call today. This page describes what it does and how to get early access.
Scanna has built a proprietary multi-agent simulation engine. Given a scenario — a press release, news article, policy draft, or counterfactual — it spins up a synthetic population of agents that post, argue, and trade across a simulated social layer and prediction market, then produces an analysis of how the event would likely play out.
How it works
The pipeline extracts entities from the scenario into a knowledge graph, generates grounded personas, runs a concurrent multi-round simulation across a social layer and a prediction market, and writes a report citing the simulated activity. Outputs are designed as data exports — a trajectory file, a reproducibility fingerprint, and a Markdown report — meant to feed back into Scanna's analytics rather than a chat UI.
Status
End-to-end runs have been validated internally. The engine is in private beta today; public access opens once it's deployed behind a public API. Want early access for research or validation? Reach out.