Tier-1 architecture slide
Agent Inventory & Dependency Explorer
A governed enterprise runtime matrix for agent purpose, autonomy, dependencies, controls, latency, resilience, identity, and approval boundaries.
Platform capability proof: every agent dependency shows the evidence used, source mode, policy gate, replay record, allowed action, blocked action, collaboration owner, observability trace, value impact, and learning boundary.
Provenance: reference dependency matrix, not live registry coverage. Use Agent Inventory and Agentic Workforce for live agent rows until this matrix is bound to AgentDefinition and runtime execution records.
Executive View
| Agent | Business Purpose | Decision Type | Autonomy Level | Real-Time Required | Latency Target |
|---|---|---|---|---|---|
| Spend Intelligence Agent | Budget protection | Probabilistic | Recommend | No | <15 min |
| Invoice Validation Agent | Invoice processing | Deterministic + Probabilistic | Governed Action | Yes | <5 sec |
| Duplicate Invoice Detection Agent | Fraud & leakage prevention | Deterministic | Autonomous | Yes | <2 sec |
| Credit Risk Agent | Counterparty exposure management | Probabilistic | Recommend + Escalate | Yes | <10 sec |
| Journal Intelligence Agent | Journal validation | Deterministic | Governed Action | No | <30 min |
| Close Intelligence Agent | Close acceleration | Probabilistic | Recommend | No | Hourly |
| Treasury Liquidity Agent | Cash forecasting | Probabilistic | Recommend | Yes | <1 min |
| Payment Risk Agent | Payment protection | Deterministic + Probabilistic | Escalate | Yes | <5 sec |
| Vendor Risk Agent | Supplier monitoring | Probabilistic | Recommend | Near Real-Time | <15 min |
| Contract Compliance Agent | Contract validation | Deterministic | Governed Action | No | <30 min |
| Forecast Intelligence Agent | FP&A forecasting | Probabilistic | Recommend | No | Daily |
| Working Capital Agent | Cash optimization | Probabilistic | Recommend | No | Hourly |
| Intercompany Agent | Intercompany reconciliation | Deterministic | Governed Action | No | Hourly |
| Tax Compliance Agent | Tax monitoring | Deterministic | Governed Action | No | Daily |
| Control Monitoring Agent | Continuous controls | Deterministic | Autonomous Detection | Yes | <5 sec |
Detailed Agent Inventory
| Agent | Mission | Decision Type | Inputs | Outputs | Source Systems | Policies | Human Approval | Latency |
|---|---|---|---|---|---|---|---|---|
| Invoice Validation Agent | Validate invoice before posting | Hybrid | Invoice, PO, GRN, Vendor Master | Approve, Hold, Escalate | SAP ECC, S/4HANA, Ariba | Invoice Policy, Approval Matrix | Above threshold | <5 sec |
| Credit Risk Agent | Prevent exposure breaches | Probabilistic | Exposure, credit ratings, payment history | Credit Hold, Escalation | SAP, Treasury, External Ratings | Credit Authority Matrix | High-risk decisions | <10 sec |
| Journal Intelligence Agent | Prevent journal errors | Deterministic | Journal, GL, Cost Center | Approve, Reject, Recommend | SAP ECC, S/4 | Accounting Policy | Required | <30 min |
| Treasury Liquidity Agent | Optimize cash position | Probabilistic | Bank balances, forecasts, payments | Cash recommendations | Treasury Platform, SAP | Liquidity Policy | Required | <1 min |
BP Sphere Runtime Metadata
- Agent Name
- Agent Version
- Business Capability
- Business Owner
- Technology Owner
- Data Owner
- Source Systems
- APIs Used
- Events Consumed
- Events Produced
- Deterministic / Probabilistic
- Confidence Score
- Policy Sets Applied
- Evidence Sources
- Human Approval Required
- Write Access
- Read Access
- PII Access
- Financial Impact
- SLA
- Latency Requirement
- Recovery Time Objective
- Replay Supported
- Learning Enabled
- Autonomous Level
- Model Used
- Prompt Version
- Cost Per Execution
- Monitoring Dashboard
- Fallback Mode
Dependency Topology
Agent
v
Events
v
Policies
v
Evidence
v
Decision
v
Human Approval
v
SAP / S4 / CFIN Writeback
Additional BP Technology Review Attributes
| Category | Attribute |
|---|---|
| Identity | Entra ID Integration |
| Identity | Service Principal |
| Identity | Agent Identity |
| Identity | Delegated Authority |
| Cyber | Zero Trust Compliant |
| Cyber | Secrets Management |
| Cyber | Encryption |
| Cyber | Network Boundary |
| Cyber | Data Residency |
| Cyber | Audit Logging |
| Data | UDP Source |
| Data | Databricks Tables |
| Data | OneData Mapping |
| Data | Data Quality Score |
| Data | Data Lineage |
| Governance | AI Risk Classification |
| Governance | Model Approval Status |
| Governance | Policy Version |
| Governance | Control Mapping |
| Governance | Regulatory Mapping |
| Runtime | Deterministic Replay |
| Runtime | Recovery Strategy |
| Runtime | Human Fallback |
| Runtime | Escalation Path |
| Runtime | Parallel Run Support |
Runtime Proof