Agent Monitor
Decision execution pipeline — simulate → decide → act → prove
Mode: LIVE · Data: LIVE_SOR
Role: Administrator
Execution Source: CHAIN
P2P live incident context · Agent Monitor
INV-000016-215 · Vendor ARIBA-16 · $11.7K
Ariba invoice not matched to approved receipt state
This surface is anchored to the same live duplicate-settlement incident as the P2P mission page. $11.7K is the duplicate-linked value on this incident. $8.8M is the replayable duplicate/exception value traced across the last 90 days. 1 similar incidents already exist for this vendor in governed tenant memory.
Agent Execution Proof
Refreshed: 2026-07-03 19:42:58 UTC
SOURCE
bp Sphere AgentExecution
Scope: Single decision execution
POLICY
Policy gates applied per agent step
EVIDENCE
Evidence pack d0a0de5d-fc08-4794-ac20-38d23a8a304c
RECORDS
Execution 5a39c7a7-3b14-4aad-8f39-2e1868a6051a
ACTION
Inspect agent steps or replay decision
Advisory only — partial execution chain available.
Decision Header
Decision ID
5a39c7a7-3b14-4aad-8f39-2e1868a6051aMission
p2p.auto_validate
Persona
Administrator
Autonomy Attempted
AL3
Autonomy Granted
AL3
Policy Version
pol:finance:control
Model Used
gpt-4o-mini
Confidence
0.99
Agent Runtime Health
CPU
N/A
Memory
N/A
Queue Depth
0
Retries
0
Drift Status
0.00
Last Failure
—
Transaction Stream
Live and historical governed transactions surfaced from the current incident scope and recent agent activity.
| Time | Transaction | Status | Risk | Decision |
|---|---|---|---|---|
| 19:42:58 UTC | Matched | Low | Auto-approved | |
| 19:42:34 UTC | Matched | Low | Auto-approved | |
| 19:42:21 UTC | Matched | Low | Auto-approved | |
| 19:42:01 UTC | Matched | Low | Auto-approved | |
| 19:41:43 UTC | Matched | Low | Auto-approved | |
| 19:41:21 UTC | Matched | Low | Auto-approved |
Deterministic Evidence
Inputs Hash
a269f78ff714ae02e5d63d786bc654b73ff8210329e5bf30133a9e4353304503Outputs Hash
79460d853e47014e72f7826ec1dd2be882584fd30ca5bad40903907e56482348Policy Evaluation Hash
26adebca9a99f27221045a6e8b0c737e651f423b4b4269ca40d4551ab3e827b8Execution Replay Hash
96ede9b1240c29b7966e0d3f7ffce5c1cdf5371d0908ba29e3327f26e8a11a93SOR Snapshot ID
snapshot.unavailableEvidence Pack
d0a0de5d-fc08-4794-ac20-38d23a8a304cScenario
Decision paused — policy gate requires human judgment.
Policy gate transparency enabled for audit review.
Execution Pipeline
| Phase | Activities | Latency | Confidence | Autonomy | Policy Result |
|---|---|---|---|---|---|
| SIMULATE | Policy Pre-check, Risk Scoring | 1ms | 0.99 | AL3 | PASS |
| DECIDE | Model Reasoning, Confidence Scoring | 1ms | 0.99 | AL3 | PASS |
| ACT | SOR Write Attempt, Policy Gate | 1ms | 0.99 | AL3 | PASS |
| PROVE | Evidence Stored, Value Attribution | 1ms | 1.00 | AL3 | PASS |
Execution Timeline
Invoice Intelligence Agent
AL3
Stage: ANALYZE · Status: Executed
Latency: 2ms · Confidence: 0.99
Model Governance
Model Version: gpt-4o-mini
Prompt Template Version: prompt.v1
Embedding Index Version: embed.v1
Token Usage: 0
Cost per Decision: $0.0000
Fail Rate (24h): 0.00%
Shadow vs Live Delta
Shadow decision: Escalate
Live decision: Approve
Delta reason: Policy override - vendor risk classification
Value Attribution
Exceptions prevented: 1
Manual hours saved: 0.6
Risk avoided: $68,200.00
Cost per decision: $0.0000
ROI ratio: 68200000.0x
Catalog Agents Available (Not executed in current window)
Policy Enforcement AgentTreasury Policy AgentPolicy Compliance Violation AgentPolicy Override MonitorLimit Breach Policy Agent