Design each configuration on the map, then capture it. Every option replaces the chosen side's units and doctrine with what is on the map at capture time; the scenario as loaded is the baseline.
Advanced: edit options JSON
Search compositions within a value budget. Units are placed around the anchor (click “set anchor”, then the map). Successive halving spends more runs on the promising designs.
| Platform | Value | Min | Max | Orders |
|---|
The on-prem LLM reads the scenario on the map, retrieves doctrine from the Quartz knowledge base and drafts courses of action; the simulator evaluates every option on common seeds. The LLM never supplies numbers: every figure in the briefing is checked against simulator output. Nothing is saved until you approve an option.
Describe the event as a news report would. Synthetic citizens from each selected side react in public and in private; BRIVA only relays the event. Draft it from the last simulation, then edit freely.
Place the assessed real-world picture on the Scenario map (or load it as a scenario), then match. The closest library scenarios are shown with the recommendations saved for them. Evaluate options on the actual picture before acting on a nearest match.
| Kind | Id | Name | Country | Status | Source | Origin |
|---|
| Id | Kind | Status | Submitted | Finished | Error |
|---|