Organize research library and schemas

papers/finance-automation/automated-finance-architecture.md

@@ -0,0 +1,26 @@
+# Automated Finance Architecture
+
+BlackRoad OS models finance automation through orchestrated agents that mirror traditional corporate finance controls while remaining auditable via PS-SHA∞ journaling. The architecture centers on eight core finance agents coordinated by a finance orchestrator and exposed through API and Prism views.
+
+## Core Components
+
+- **Finance orchestrator:** Routes tasks, enforces segregation of duties, and coordinates PS-SHA∞ logging.
+- **Core agents:** Close, treasury, FP&A, procurement, compliance, tax, reporting, and controls agents handle domain-specific actions.
+- **API + Prism views:** External interfaces for ingesting events, triggering workflows, and surfacing journaled outcomes.
+
+## Reference Foundations
+
+- **Corporate finance literature:** Valuation, risk, and capital structure concepts from Brealey and OpenStax guide agent policies and scenario modeling.
+- **Regulatory expectations:** FINRA/SEC focuses on suitability, surveillance, and market integrity influence orchestration rules.
+- **Automation handbooks:** Industrial automation patterns inform safety constraints, redundancy, and recovery strategies.
+
+## PS-SHA∞ Integration
+
+- Every financial action is journaled with hashes and prior links to preserve an auditable worldline.
+- Contradictions between agents (e.g., revenue vs. compliance) are recorded with resolution metadata.
+- Journal entries can be exported for external attestations or compliance tooling.
+
+## TODOs
+
+- Elaborate data contracts for each agent API and align with `schemas/journal-entry.schema.json`.
+- Prototype scenario simulations to stress-test segregation-of-duties policies.