Dashboard-first mission control for secure, coalition-aware intelligence operations. Gotham resolves investigations and entity tracking, Foundry governs ontology and pipelines, AIP coordinates copilots and agent evals, and Apollo controls deployment rings, rollback, and runtime policy.
Gotham resolves investigations and entity tracking; Foundry owns ontology, pipelines, lineage, and application logic; AIP runs copilots, agents, evals, and workflow automation; Apollo controls deployment rings, rollback, and runtime policy.
Operator feedback, corrections, query logs, alert outcomes, and mission results become eval datasets. Suggested prompt, workflow, heuristic, and routing upgrades remain staged until human approval and audit capture.
Need-to-know authorization, row/column/entity permissions, coalition compartments, provenance controls, immutable logs, and policy-as-code protect every data read, tool call, agent handoff, and operational package.
Representative backend workers emit typed events, score confidence, run drift checks, and propose safe upgrades with rollback metadata.
def propose_upgrade(feedback, evals):
if evals.precision < .92 or feedback.risk == 'high':
return {'state':'review','gate':'human'}
return {'state':'staged','rollback':'apollo:ring-1'}