Inject behavioral shocks into the live sim. DRO re-resolves all downstream agent decisions in roughly 4.2s per branch.
When DRO opens a scenario, the engine instantiates a curated cast of personas — drawn from a 1,184-deep library of fitted profiles built on a decade of public posting history, voting records, draft-night quotes, and beat-by-beat reaction data. Each persona carries its own tone, posting cadence, source network, bias dials, and platform-native voice.
An event fires. Personas observe what's relevant to them, post in character, then read each other and update their priors. Aggregate signals (sentiment, narrative dominance, topic volume) feed back into rival GM agents — who incorporate the room's reaction into their next decision. Then the cycle repeats.
Every report your shop has written — region scouts, area scouts, cross-checkers, national. Tagged by author, revision, conviction. The agents inherit your grading scale, not a generic one.
Indianapolis pulls. Pro day re-tests. In-season GPS load. Force-plate asymmetry. Sleep and HRV when available. The radar that drives the radar.
Public draft & transfer histories of every rival GM and Sporting Director — fitted into a behavioral profile. Risk tolerance, cap posture, owner relationship, contract leverage, media discipline.