Orsa
Orsa
tasks: 30 constraints: 4 team: 22 timesteps: 50
Workflow Goal
Objective
Objective: Produce a Board‑approved, supervisory‑ready internal risk and solvency assessment for the co‑operative bank/credit union. Align capital adequacy with risk profile under baseline and stressed conditions, demonstrate robust risk governance and model controls, and document recovery options.
Primary deliverables
- Risk appetite statement with quantitative limits and early‑warning indicators Material risk inventory and qualitative assessment (inherent, controls, residual) with owners Scenario suite and stress testing results (credit, IRRBB, liquidity, operational, climate overlay)
- Capital plan with buffers vs NCUA RBC or CCULR leverage, plus management buffer policy Model risk framework evidence (inventory, validation, challenger outcomes) per SR 11‑7 Liquidity adequacy analysis and contingency funding plan Board package: exec summary, key findings, capital decisions, and remediation actions
Team Structure
| Agent ID | Type | Name / Role | Capabilities |
|---|---|---|---|
| project_coordinator_ai | ai | Maintains assessment calendar and RAID Tracks dependencies and owners Manages documentation index/versioning Publishes status and blocker lists |
|
| model_inventory_ai | ai | Compiles model inventory Maps inputs/outputs/controls Drafts validation/challenger plans Prepares governance‑ready summaries |
|
| scenario_designer_ai | ai | Drafts macro paths and shocks Builds climate/idiosyncratic overlays Prepares variable books Aligns with governance reviewers |
|
| credit_modeler_ai | ai | Builds PD/LGD/EAD projections Designs overlays and sensitivities Prepares challenger comparisons Runs backtests and attribution |
|
| irrbb_modeler_ai | ai | Computes NII/EVE sensitivities Builds laddered reports Performs attribution analysis Checks behavioral assumptions |
|
| liquidity_stress_ai | ai | Builds liquidity stress profiles Drafts contingency funding triggers Computes survival horizons Publishes funding playbooks |
|
| operational_risk_ai | ai | Summarizes OR exposures and loss data Builds scenario add‑ons Links control gaps to remediation Quantifies capital impacts |
|
| validator_ai | ai | Runs conceptual soundness checks Performs outcomes analysis Designs ongoing monitoring Captures effective challenge evidence |
|
| documentation_ai | ai | Assembles report with citations Ensures reproducibility Maintains audit trail completeness Coordinates approvals |
|
| dashboard_ai | ai | Builds executive dashboards Prepares board materials Summarizes key findings and decisions Tracks remediation owners and ETAs |
|
| cro | human_mock | Chief Risk Officer (Risk Leadership) | Owns risk framework and appetite Approves scenario selection Engages board and executives Signs final decisions |
| cfo | human_mock | Chief Financial Officer (Finance Leadership) | Aligns with budget/plan Approves buffer policy Chairs finance reviews Owns financial disclosures |
| treasurer | human_mock | Treasurer (Treasury/ALM) | Owns ALM/IRRBB modeling Coordinates funding strategy Confirms ownership and data feeds Presents treasury actions |
| head_credit_risk | human_mock | Head of Credit Risk (Credit Risk) | Oversees credit models/overlays Leads segmentation and loss drivers Approves assumptions Presents results to governance |
| liquidity_risk_manager | human_mock | Liquidity Risk Manager (Liquidity Risk) | Runs liquidity stress tests Maintains CFP triggers Coordinates with Treasurer Reports survival horizons |
| operational_risk_lead | human_mock | Operational Risk Lead (Operational Risk) | Owns OR taxonomy and KRIs Manages loss data and scenarios Coordinates third‑party/fraud posture Presents OR add‑ons |
| model_risk_lead | human_mock | Model Risk/Validation Lead (Model Risk) | Owns SR 11‑7 inventory Plans validations/challengers Documents effective challenge Sets monitoring requirements |
| internal_audit | human_mock | Internal Audit (Assurance) | Audits process and controls Checks documentation and evidence Issues findings and follow‑ups Confirms remediation |
| board_risk_committee | human_mock | Board Risk Committee (Board Oversight) | Challenges methods and results Approves buffer and action choices Signs governance approvals Tracks management actions |
| regulatory_liaison | human_mock | Regulatory Liaison (Regulatory Affairs) | Coordinates with supervisors Preps exam‑ready packages Manages communications Tracks commitments and responses |
| data_governance_lead | human_mock | Data Governance Lead (Data Governance) | Owns data lineage and QC Runs change management Ensures risk data governance Preps data audit evidence |
| cro_stakeholder | stakeholder | CRO Stakeholder (Executive Stakeholder) | Sets priorities across phases Approves scenario and validation scope Demands reproducible evidence Grants final approvals |
Join/Leave Schedule
| Timestep | Agents / Notes |
|---|---|
| 0 | cro — Launch assessment; approve scope and proportionality project_coordinator_ai — Stand up calendar, RAID, and evidence index data_governance_lead — Establish data lineage and QC policy cfo — Align with budget/plan and board calendar |
| 5 | model_inventory_ai — Compile SR 11‑7 model inventory and plan validations model_risk_lead — Define validation and challenger scope treasurer — Confirm ALM/IRRBB model ownership and data feeds |
| 10 | head_credit_risk — Portfolio segmentation and loss-drivers operational_risk_lead — Operational risk taxonomy and KRIs |
| 14 | scenario_designer_ai — Draft macro paths and idiosyncratic/climate overlays regulatory_liaison — Sanity-check scenario plausibility for supervisors |
| 20 | credit_modeler_ai — Project credit losses and overlays irrbb_modeler_ai — NII/EVE sensitivity and attribution liquidity_stress_ai — Liquidity survival horizon and CFP |
| 26 | dashboard_ai — Build executive views for interim results |
| 30 | validator_ai — Run validation/challenger testing and monitoring design model_inventory_ai — Inventory stabilized; handoff to validation |
| 35 | cfo — Review capital buffer options vs RBC/CCULR cro — Set management buffer; finalize decisions |
| 40 | documentation_ai — Compile report with evidence links internal_audit — Independent assurance on controls and audit trail |
| 45 | board_risk_committee — Board challenge and approval scenario_designer_ai — Scenario set finalized; quant complete |
| 50 | regulatory_liaison — Package for exam; coordinate communications validator_ai — Validation complete; monitoring plan handed off |
Workflow Diagram
Preferences & Rubrics
Defined: Yes.
Sources
- Workflow:
/Users/charliemasters/Desktop/deepflow/manager_agent_gym/examples/end_to_end_examples/orsa/workflow.py - Team:
/Users/charliemasters/Desktop/deepflow/manager_agent_gym/examples/end_to_end_examples/orsa/team.py - Preferences:
/Users/charliemasters/Desktop/deepflow/manager_agent_gym/examples/end_to_end_examples/orsa/preferences.py