Phase 1 - Immediate foundation

Enterprise Finance Ontology Manager

Create the decision ontology that connects objectives, KPIs, missions, decisions, policies, events, entities, evidence, actions, outcomes, agents, and learning into one finance intelligence layer.

Current gap: entity-centric ontology creates navigation; decision ontology creates intelligence. Open decision ontology JSON Runtime contract Runtime coverage Governance report Runtime chain JSON Back to foundations

Runtime Hardening Proof

Execution evidence from Enterprise Decision Runtime service

This panel is generated from runtime service functions on each request: event correlation, policy simulation, learning metrics, agent guardrails, decision trace, and audit package export. It closes the view-vs-system gap by showing the platform executing, not only describing architecture.

7registered agents

5registered policies

1correlated signal(s) from 3 events

$450.0Kpolicy simulation amount

100.0%human acceptance over 2785615 decided executions (90d)

5825088decision memory records

BLOCKbad-data guardrail result

route_to_human_supervisoragent failure fallback

Live event correlation

SignalPriorityImpactEvent IDs

sig-2efe089ec8P1$1.4Mevt-001, evt-002, evt-003

Policy + guardrail hardening

Control	Runtime value
Policy	human-approval-policy
Prior threshold	$50.0K
Simulated threshold	$100.0K
Outcome	escalate
Approval volume change	-18%
Guardrail failures	missing_evidence, low_source_confidence, writeback_requires_human_approval
Hallucination guard	evidence_required_before_recommendation

Replay + audit export

Artifact	Value
Audit package	audit-97ab17991395
Integrity hash	c80fd1aea51cec3d76a8d2e6346938acd1938531f544f8f3f7fd3917f1d6c98c
Regulatory export ready	True
Trace steps	10
Runtime chain	Signal -> Context -> Ontology -> Policy -> Evidence -> Decision -> Governance -> Action -> Replay -> Learning

Runtime APIs

Endpoint
/api/finance-foundations/runtime-proof
/api/finance-foundations/enterprise-finance-ontology-manager/runtime-proof
/api/iaf/enterprise-decision-runtime/learning-metrics
/api/iaf/enterprise-decision-runtime/agent-guardrails
/api/iaf/enterprise-decision-runtime/policy-simulation
/api/iaf/enterprise-decision-runtime/event-correlation

Hardening stance: agents fail closed to human supervision on missing evidence, low source confidence, or write-back attempts. Learning is measured, replay-backed, and cannot change policy or autonomy without approval.

Open Enterprise Decision Runtime Open runtime proof JSON

North Star shift

Entity-CentricVendor -> Invoice -> Journal. Useful for navigation, insufficient for decision intelligence.

Decision-CentricObjective -> KPI -> Decision -> Policy -> Evidence -> Action -> Outcome -> Learning.

Runtime ImpactEvery signal resolves to what decision is required, why, by whom, under which policy, with what proof.

Architecture challenge answerThe system knows what to do because the ontology encodes decisions, actions, outcomes, and learning.

Agent BoundaryAgents act only on ontology-declared entities, policies, skills, permissions, and actions.

Audit BoundaryReplay follows the ontology chain from objective through outcome and learning.

Recommended 11-layer finance decision ontology

Layer	Name	Purpose	First-class entities	Example
L1	Financial Objectives	Explains why work exists and what enterprise outcome matters.	Cash Optimization, Working Capital, Risk Reduction, Compliance, Margin Protection, Close Acceleration ...	Duplicate invoice prevention maps to Cash Optimization, Risk Reduction, and Compliance.
L2	KPI Ontology	Makes performance measures first-class objects so decisions can be simulated and measured.	DSO, DPO, Cash, Working Capital, Close Readiness, Forecast Accuracy ...	Payment hold impacts cash protected, duplicate exposure, DPO, and control effectiveness.
L3	Finance Mission Ontology	Organizes processes, decisions, KPIs, policies, agents, and signals by finance mission.	P2P, O2C, R2R, FP&A, Treasury, Tax ...	Invoice risk belongs to P2P but also impacts Treasury cash and Controls assurance.
L4	Decision Ontology	Maps every signal to the decision a human or agent must form.	Approve Invoice, Release Payment, Escalate Credit Risk, Post Journal, Approve Spend, Release Cash ...	Duplicate invoice risk maps to the decision: trigger payment hold or approve release.
L5	Policy and Control Ontology	Turns policies into governed objects that explain what is allowed and who must approve.	Policy, Control, Approval Authority, Materiality Threshold, Risk Rating, Delegation Authority ...	Three-way match policy and duplicate payment control govern the invoice release decision.
L6	Event Ontology	Makes finance dynamic by treating operational change as decision triggers.	Invoice Received, Invoice Approved, Invoice Rejected, Payment Released, Credit Downgrade, Shipment Delayed ...	Invoice received plus duplicate pattern triggers the duplicate invoice risk signal.
L7	Finance Entity Ontology	Anchors source-system objects and relationships used to resolve context.	Vendor, Customer, Contract, Purchase Order, Invoice, Journal ...	Supplier, contract, PO, invoice, goods receipt, and prior payment history are resolved as one context chain.
L8	Evidence Ontology	Makes audit proof inspectable before recommendations are trusted.	Evidence Item, Evidence Pack, Source Record, Supporting Document, Approval Record, Audit Record ...	Decision links to invoice image, PO, goods receipt, supplier master, approval record, and replay trace.
L9	Action Ontology	Defines what can actually be done and which actions are allowed, drafted, denied, or escalated.	Approve, Reject, Escalate, Route, Create Journal, Create Case ...	Agent may draft a payment hold, but payment release remains denied unless certified for controlled autonomy.
L10	Outcome and Value Ontology	Connects actions to realized results and value attribution.	Fraud Prevented, Cash Collected, Close Accelerated, Exposure Reduced, Risk Avoided, Value Realized ...	Invoice hold results in duplicate payment prevented and cash protected.
L11	Agent and Memory Ontology	Shows which agents, skills, tools, permissions, outcomes, patterns, and learning objects improve future decisions.	Agent, Skill, Tool, Capability, Permission, Autonomy Level ...	Duplicate Invoice Agent uses matching, evidence, and policy skills; outcome becomes a reusable exception pattern.

Example: duplicate invoice risk decision chain

ObjectiveCash Optimization + Risk Reduction + Compliance

KPICash protected, duplicate exposure, DPO, control effectiveness

MissionP2P primary, Treasury and Controls secondary

ProcessInvoice validation before payment release

DecisionTrigger payment hold or approve release

PolicyThree-way match, duplicate payment control, authority boundary

EventInvoice received with same supplier, amount, and PO pattern

EntitySupplier -> Contract -> PO -> Goods Receipt -> Invoice -> Payment

EvidenceInvoice image, PO, GR, prior payment, approval record, source lineage

ActionDraft payment hold, request evidence, escalate to supervisor

OutcomeDuplicate payment prevented, cash protected, control exception closed

LearningPattern stored for future duplicate-risk detection and replay

Credibility coverage tests

Coverage	Rule	Target	Why it matters
Decision coverage	Every signal must map to an explicit decision	Target 100%	Prevents agents becoming report generators
Policy coverage	Every recommendation must cite policy/control objects	Target 100%	Makes decisions explainable and auditable
Evidence coverage	Every material decision must create an evidence pack	Target 100%	Supports audit and replay
KPI coverage	Every decision should map to one or more KPIs	Target 95%	Enables simulation and value attribution
Outcome coverage	Every action should track realized outcome	Target 95%	Enables learning and value realization
Agent coverage	Every decision should show participating agents/skills/tools	Target 100%	Makes agentic work credible

Core components

Component	Capability
Decision Ontology	Every signal maps to a finance decision such as approve invoice, release payment, escalate credit risk, post journal, close period, or release cash.
Policy Ontology	Policies, controls, approval authority, materiality thresholds, risk ratings, delegation authority, and compliance requirements are first-class objects.
Event Ontology	Invoice received, journal posted, payment released, credit downgrade, FX exposure breach, cash threshold breach, and close exception events activate decisions.
Action Ontology	Approve, reject, escalate, route, create journal, create case, request evidence, initiate collection, trigger hold, and create accrual are explicit action objects.
Outcome and Value Ontology	Actions resolve to outcomes such as fraud prevented, cash collected, close accelerated, exposure reduced, DSO improved, and value realized.
Entity Registry	Vendor, Customer, Contract, Purchase Order, Invoice, Journal, Cost Center, Asset, Business Unit, Legal Entity, Bank Account, Commodity Position.
Relationship Manager	Objective -> KPI -> Mission -> Process -> Decision -> Policy -> Event -> Entity -> Evidence -> Action -> Outcome -> Learning.
Ontology Explorer UI	Show all invoices for supplier, all journals from contract, all decisions impacting an asset, and all KPIs impacted by a decision.
Ontology APIs	Entity, relationship, and context retrieval APIs for agents and UI drilldowns.
Governance	Version control, approval workflow, change history, and impact analysis.

Agent family

Agent	Purpose
Ontology Discovery Agent	Finds candidate objects and missing entity coverage.
Relationship Discovery Agent	Discovers cross-system object links and confidence.
Ontology Steward Agent	Routes changes to data owners and records approval history.
Decision Mapping Agent	Maps signals and entities to decisions, policies, actions, KPIs, objectives, and replay requirements.
Outcome Learning Agent	Links decisions to realized outcomes and reusable learning patterns.

Proof links

Surface	Route
Decision Ontology JSON	Open
Runtime Contract	Open
Runtime Coverage	Open
Governance Report	Open
Runtime Chain	Open
Hybrid Finance Simulator	Open
Context Graph	Open
Ontology Explorer	Open
Object Journey	Open
Data Integrations	Open

API contract

Endpoint
/api/finance-foundations/enterprise-finance-ontology-manager
/api/finance-foundations/ontology/decision-model
/api/iaf/ontology/decision-ontology/contract
/api/iaf/ontology/decision-ontology/coverage
/api/iaf/ontology/decision-ontology/governance
/api/iaf/ontology/decision-ontology/chain/{signal_id}
/api/iaf/ontology/decision-ontology/backfill
/api/finance-foundations/ontology/entity/{id}
/api/finance-foundations/ontology/context/{id}

Validation questions

Question
Which decision does this signal require?
Which policy, evidence, action, KPI, objective, agent, and learning pattern are linked?
Show all invoices for a supplier.
Show all journals created from one contract.
Show all decisions impacting an asset.
What changed in the ontology and who approved it?