P2P Analyst Workspace
Oversight workspace for a blended human + agent team
SOR Data21 Jul 2026 10:31 UTC·Partial (2/4 live)·Sources: 4·85 SORs·Recon: Active·Canonical Model
Route: /ui/mission/p2pMission: p2pRole: administratorData: PARTIAL (2/4 live)Audit Mode: OFFDecision Access: Page → Queue Drawer → Case Drawer → Evidence / Execution
P2P Command Center
P2P Mission · Health · Actions · Risks
P2P Administrator proof layer
Operations, learning, memory, outcome, and trust are visible on this role surface
This strip keeps non-analyst P2P pages enterprise-grade: team backlog, SLA risk, escalation queue, policy recommendations, automation candidates, knowledge gaps, source-system lineage, confidence, Ask Sphere access, and outcome attribution remain visible without duplicating the analyst workbench.
Truth tier: T4_LIVE_SOR_LINEAGEAttestation: eligibleSource repair: no
Team operations
8 backlog items
SLA risk, escalations, workload, bottlenecks, and capacity are tracked for supervisor action.
Decision confidence
$5.9M
Priority is ranked by value, risk, SLA, policy severity, evidence readiness, and financial impact.
Learning queue
Governed
New patterns, policy recommendations, automation candidates, and knowledge gaps require steward approval before reuse.
Enterprise memory
Traceable
Similar cases, prior outcomes, expert guidance, and reuse recommendations are linked to replay-backed evidence.
Outcome attribution
Tracked
Agent, human, policy, financial, and learning contributions are separated so value claims stay explainable.
Ask Sphere
Available
Role-level questions can open portfolio, policy, supplier, evidence, lineage, replay, and supervisor handoff answers.
Supervisor learning stewardship actions
Learning actions remain governed. Approve, reject, promote, retire, and backtest decisions are recorded through the learning governance runtime when the decision-memory script is present; otherwise this opens the decision memory record for review.
No learning governance action submitted in this browser session.
Trust signals: Evidence available · Source system visible · Freshness visible · Confidence visible · Policy visible · Replay available · Human approvals visible · Authority limits visible · Audit trail visible.
Governed Runtime Surface · P2P
From Signal -> Decision -> Action -> Proof
The runtime is operating inside policy right now. No governed payment decisions require judgment, but the proof, replay, and system-pulse rails remain available.
Event FabricAgentic FabricEvidence & Replay
Decision Support And System Pulse
Mission-wide context stays available, but it no longer competes with the governed queue above.
Governed decisions that need you now
Routine invoice flow was absorbed automatically. These exceptions stopped because policy, evidence, duplicate-risk, or authority boundaries now need human judgment.
0 governed decisions remain$0 awaiting judgmentLatest touch 10d 11h agoAutonomy coverage 0%
Mission runtimeOperational intelligenceLead command centerBusiness unitsCFO / treasuryCompact default
Current live threat
INV-000016-215 · $12K
Vendor ARIBA-16 · Ariba invoice not matched to approved receipt state
Open decision drawerWhat needs attention now
Live transaction stream
5 governed executions
No autonomous clears are recorded in the current execution window yet.
Active agent fabric
1 agents online
0 routed to governed review · 0 escalated across the last 24h.
Mission queue pressure
0 decisions awaiting judgment
$0 remains queue-bounded behind policy, evidence, or authority checks; this is separate from duplicate-linked value already contained.
SOR connectivity
Live source labels pending
Canonical tenant scope is live even if source labels are not populated in this render.
Full live SOR agentic fabric
This mission is running in full live SOR realism mode, without compact validation floor or non-attested fallback.
Financial Integrity Agent
WATCHING
Duplicate and payment-release anomalies are gating 0 known items.
Vendor Supplier and Payment Risk Agent
IDLE
Vendor pattern, bank-account, and authority changes are attached to governed exceptions before release.
Policy Compliance Agent
WATCHING
0 governed checkpoints were created only after policy, authority, or evidence boundaries were crossed.
Settlement Exposure Agent
WATCHING
Settlement exposure currently binds $0 of duplicate-linked value.
Treasury Financial Impact Simulation Agent
WATCHING
Cash and settlement posture is tracking $0 of live invoice value awaiting judgment.
Evidence Pack Assembly Agent
SEALED
20 recent governed decisions already carry evidence or replay lineage.
Supervisor Agent
STANDBY
0 escalations were routed onward when analyst authority or SLA boundaries were crossed.
Human Team Collaboration and Handoff Agent
STANDBY
0 governed cases remain routed for human judgment, with 0 already pushed beyond the analyst boundary.
Operational impact
Agents absorb routine flow first. The analyst intervenes only when policy, evidence, duplicate risk, or authority boundaries require governed judgment.
Live queue and execution data
Previous role
Manual AP handling · 6.2h/day
Manual validation55%
Mismatch investigation15%
Supplier contact10%
Admin tracking20%
Current governed role
Judgment and exception control · 1.9h/day
Exception judgement70%
Risk validation20%
Strategic negotiation10%
Autonomous executions, governed decisions, and duplicate protection are derived from the live 24h operating window. Explicit financial impact only counts decisions that recorded it directly. Linked invoice value under review stays separate so the page does not imply realized savings. Explicit financial impact is only shown when a governed decision row actually recorded it.
Operating context
Data as of: 2026-07-20 22:11 UTC (canonical ingest)Last live SOR sync: 2026-07-12 17:56 UTCInvoices in scope: 1,819 (tenant scope)Value in scope: $162.8M (modelled · iaf_invoices.net_amount)Source systems: 11 contributingEvidence: 0 packs attachedPolicies: 0 active controlsReplay: 0 decision paths available
Operational impact
Protected exposure, bounded release pressure, and governed containment quality stay visible after the queue.
Exposure prevented
$0
Queue-bounded value awaiting judgment
$0
Replayable duplicate/exception value
$8.8M
Autonomous resolution rate
0%
Escalation avoidance
0%
Evidence-backed decision rate
100%
Value generated by agentic fabric
Duplicate Invoice Detection Agent
$0
Vendor Supplier and Payment Risk Agent
1 current-vendor anomalies in replayable history
Treasury Financial Impact Simulation Agent
$0
Evidence Pack Assembly Agent
20 replay-ready governed decisions
Decision intelligence
Why the recommended path is weighted, explicit, and evidence-backed.
Prior vendor anomalies
1 in live replay memory
Duplicate / exception replay similarity
80 similar incidents
Policy checkpoint demand
0%
Treasury review pressure
$0
Supervisor override history
0%
Runtime execution and ontology
Invoice -> Vendor -> PO -> Contract -> Cost Center -> Payment Risk
The runtime resolves relationships, prior incidents, settlement posture, and authority boundaries before payment release is allowed.
1. Signal received from enterprise event fabric
2. Context resolved across invoice, vendor, contract, and payment-risk history
3. Policies evaluated against duplicate-risk, authority, and treasury posture
4. Agents coordinated containment, escalation, and evidence sealing
5. Human escalation triggered only where autonomy boundaries were crossed
Orchestration is coordinated operational execution.
Evidence and replay
Each exception becomes replayable enterprise memory instead of disappearing into a workflow queue.
Duplicate containment protected $0 in the current runtime window. 20 governed decisions already carry replay and evidence lineage.
Assurance creates operational trust.
Decision WorkspaceEvidence VaultReplay CenterDecision MemoryObject JourneyAgent MonitorSimulation RuntimeLearning FabricValue Attribution
10d 11h ago
escalated to supervisor · Escalate · BP-SVC-884220
HUMAN
17d 14h ago
decision recorded · decision.recorded
HUMAN
17d 14h ago
decision recorded · decision.recorded
HUMAN
17d 14h ago
decision recorded · decision.recorded
HUMAN
Memory and precedent
80 similar duplicate/exception incidents are replayable in live tenant history.
The runtime has already seen 80 materially similar incidents across 27 supplier clusters in the last 90 days. 1 involved the same vendor as the current live threat.
INV-000016-215→Vendor ARIBA-16→80 prior incidents→0 policies triggered→0 supervisor escalations→$0 treasury impact→20 audit / replay lineage
Invoice
INV-000016-215
Ariba invoice not matched to approved receipt state
Vendor
ARIBA-16
1 replayable anomalies in 90d
Prior incidents
80 replayable cases
27 supplier clusters in scope
Policy + treasury
0 governed checkpoints
$0 awaiting governed release
Audit + learning
20 replay-ready
1% linked to this vendor cluster
Duplicate anomaly recurrence
80 replayable duplicate/exception incidents across 27 supplier clusters
Keep duplicate-settlement comparators warm and reuse the strongest prior containment path.
Vendor concentration
Current vendor cluster contributes 1% of replayable anomaly history
Raise vendor-risk pre-check priority before payment-release authority is evaluated.
Escalation pressure
0% of governed checkpoints escalated beyond the analyst boundary in the current live window
Refine authority routing and evidence completeness before the next similar exception lands.
Replay readiness
20 decisions are already replay-backed in this runtime slice
Reuse replay-backed outcomes first when recommending the next governed containment path.
Downstream impact and drift
The runtime is tracking where human dependence, escalation pressure, duplicate concentration, and treasury impact may drift out of healthy bounds.
Duplicate-linked queue pressure
0 live duplicate cases
Manual intervention dependency
0 open governed decisions
Escalation dependency
0 escalations in the last 24h
Treasury timing volatility
$0
Enterprise ripple effects
Treasury
$0 remains release-bounded while the duplicate threat is contained.
FP&A
$0 is the scenario-mode duplicate stress line for executive planning.
Supplier management
Vendor ARIBA-16 now carries 1 replayable anomalies in this cluster.
Compliance + audit
20 decisions are replay-backed and 100% of the live governed window is evidence-ready.
Working capital
$0 is currently protected from duplicate settlement release.
Simulation runtime
Scenario assumptions are derived from the current live incident window, not invented outcomes.
These bounded what-if cases project the same live threat into treasury, queue pressure, and duplicate-value risk without confusing simulation with observed truth.
Scenario: duplicate rate +18%
$0 projected duplicate-linked protection window
Assumption-driven sandbox from the current live duplicate window: 1 duplicate-linked cases instead of 0.
Scenario: current vendor cluster deteriorates
1 governed decisions if one more ARIBA-16 anomaly lands
Adds one more governed checkpoint and approximately $12K of duplicate-settlement pressure from the current live incident baseline.
Scenario: payment holds exceed current threshold
$12K treasury-linked value under review
Derived from current pending governed value plus the live incident amount; this stays a scenario assumption until another real event enters the runtime.
Real · seeded
PO compliance
illustrative
41%
→
83%
$18.0M
90% confidence
Standardize PO adoption; enforce no-PO/no-pay; agentify exceptions
Live operational moments and signals
One signal surface for what changed, why it matters, candidate transactions, policy/evidence context, and the governed response path.
Learning and memory
3 patterns updated from today’s governed P2P decisions
Duplicate detection, authority routing, and evidence thresholds are being tuned from replay-backed human decisions.
Operating Model
Systemic Risk · Bottleneck · Controlling policy · Escalation path · Event chain
Expand
Operating Model
Systemic Risk · Bottleneck · Controlling policy · Escalation path · Event chain
Mission Runtime Standard
Tier 1 Gold Standard
Contract: full
Performance: analyst_fast
Primary London workshop mission with the strongest analyst and supervisor scrutiny.
Operating Model
Shared platform contracts, mission decision flow, and tenant context.
This section keeps only the net-new operating context that does not already appear in the mission summary above.
Current bottleneck
Duplicate invoice cluster
Systemic Risk
High
Escalation required: Yes
Controlling policy
pol:procurement:policy
Policy evidence is bound to the authoritative incident thread.
Escalation path
Supervisor
Mitigation: Escalate to supervisor
Event chain
event
Invoice receivedagent
Validation agentpolicy
Policy gatehuman
Analyst reviewevidence
Evidence ledger
2/4 LIVE, 2 unavailable
|
Refreshed 10:31:58 UTC
|
All data within SLA
|
Source binding posture: 2/4 live · 2 unavailable · no synthetic substitution