RecordArc public release R21
Skip to main content
Book Private Demo

Decision Proof Infrastructure

Make every high-risk decision easier to explain, replay and defend.

RecordArc connects fragmented evidence, policy, model and rule context, human judgment and approvals into one governed, replayable decision record—without changing source systems or making decisions.

Read-only by designVendor-neutralReplayable proofNo writeback
Book Private DemoExplore the Decision Proof Lab
READ-ONLY PROOF BOUNDARYAuthority remains outside the layer
Decision Proof Layer authority boundarySource-system evidence and governing context enter the read-only Decision Proof Layer. Review evidence leaves the layer for authorized human review. No relationship writes back to the source decision path.Source-systemevidenceGoverningcontextRecordedactionsDecision ProofLayerread-onlyAuthorizedhuman review
  1. Source-system evidenceenters read-onlyDecision Proof Layer
  2. Governing contextretains versionDecision Proof Layer
  3. Recorded actionsretain authorityDecision Proof Layer
  4. Decision Proof Layerprepares evidence forAuthorized human review
No writebackSource and human decision authority remain unchanged.

WHEN A DECISION IS CHALLENGED

The evidence is rarely in one place.A regulator, auditor, or reviewer asks why. Evidence spans systems, rules, models, notes, and approvals.
  • Fragmented context
  • Manual reconstruction
  • Reliance on institutional memory

THE OPERATING BURDEN

When a decision is challenged, the evidence is rarely in one place.

More systems, AI-assisted workflows, and changing rules or models distribute context as scrutiny increases.

01Operating burden

Fragmented evidence

Decision trails split across systems, notes, and workflows.

02Operating burden

Slow reconstruction

Teams rebuild decisions before explaining or reviewing them.

03Operating burden

Lost context

Versions, rationale, and approvals become disconnected.

04Operating burden

Weak replay

Records show outcomes without the paths that produced them.

WHY EXISTING SYSTEMS PRESERVE FRAGMENTS

Every system preserves a fragment. RecordArc preserves the decision story.

Source systems process transactions, cases, rules, models, reviews, and approvals. RecordArc connects their fragments without replacing or changing them.

Central Decision Proof Layer relationshipsThree source-side contexts connect through orthogonal read-only relationships to a central Decision Proof Layer. Three review and output contexts connect on the other side. Source systems and people retain decision authority.retained evidencegoverning versionrecorded authorityassurance viewcontrol evidencereview packageSource recordsGoverning contextRecorded human actionsDecision ProofLayerAudit and assuranceControl testingAuthorized review
  1. Source recordsconnects read-only to Decision Proof LayerAudit and assurance
  2. Governing contextconnects read-only to Decision Proof LayerControl testing
  3. Recorded human actionsconnects read-only to Decision Proof LayerAuthorized review
Logical centralization of proof visibility, not mandatory physical centralization of all customer data.

Read-only. No writeback.Source systems retain recorded outcomes; people retain review authority.

THE DECISION PROOF RECORD

One governed record connects the context needed to explain what happened.

RecordArc connects evidence, policy, rule or model context, judgment, and approvals into a replayable proof record with visible gaps.

  1. 01
    Source evidenceRetained record from an approved source
  2. 02
    Policy versionDecision-time policy or control context
  3. 03
    Model and rule traceRecorded version, threshold, and signal context
  4. 04
    Human approvalAuthorized review, exception, or override context
  5. 05
    Structured proof packageConnected evidence prepared for later review
  6. 06
    Reconstruction timelineEvent, decision, capture, and later context

QUALITATIVE OPERATING OUTCOMES

From fragmented records to reviewable decision proof.

Connected proof makes retained context easier to assemble, inspect, replay, and challenge.

  1. 01ReduceManual evidence reconstruction
  2. 02ShortenAudit and review preparation effort
  3. 03AccelerateDecision replay from retained context
  4. 04ImproveEvidence consistency across systems
  5. 05LimitDependence on institutional memory

GOVERNED OUTPUTS

Inspectable as one proof structure

Source context

Source fact

Certified interpretation

Human-certified interpretation

Reconstructed decision record

RecordArc normalized

Connected evidence relationships

RecordArc derived

Evidence gaps and conditions

RecordArc derived

Decision timeline

RecordArc derived

CURRENT WEDGE AND EXPANSION

Start with transaction decisions made in real time across banking and payment flows.

Real time describes the source decision environment; RecordArc remains retrospective, read-only, and off-path.

Authentication and authorizationRisk, referral, hold, and declineOverride and operational exceptionRecorded outcomes only; RecordArc does not execute them
01

Banking & Payment Decision Proof

First wedge for reconstructing retained evidence around banking and payment transaction decisions.

Under developmentReview pathway
02

Financial Crime & Compliance

Adjacent workflows can reuse the proof pattern while retaining authority.

Intended enterprise capabilityReview pathway
03

Future Applications

Broader use requires governed evidence, interpretation, and authority.

Not currently offeredReview pathway

Current public implementation is a synthetic reference. Enterprise capability remains separately scoped.

CURRENT PUBLIC DEMONSTRATIONS

Two synthetic records make distinct proof paths inspectable.

The flagship payment and fraud-risk reconstruction and the separate AML alert-review reconstruction share the Decision Proof architecture while preserving different source workflows and relationship authority.

FLAGSHIP SYNTHETIC RECORDPayment and fraud-risk reconstruction

Inspect six retained nodes as ordered context with source identity, classifications, chronology, a human exception, recorded outcome, and no invented typed payment-edge semantics.

Separate AML example
Approved synthetic AML typed relationships connect alert-review evidence without rerunning or disposing the alert.
Shared posture
Both demonstrations are synthetic, retrospective, read-only, and off-path.
Permanent boundary
No authorization, scoring, monitoring, closure, disposition, writeback, or decision rerun.
Open Payment Decision ReconstructionView AML Example Overview

ENTERPRISE READINESS

Designed to fit regulated enterprise environments.

A bounded, read-only layer preserves source ownership, human accountability, and customer deployment control.

SOURCE AUTHORITY

Systems preserve facts and recorded outcomes.

RECORDARC

Connects evidence and exposes limitations.

HUMAN AUTHORITY

People retain judgment, approval, and review.

No writeback. No autonomous disposition. No decision replacement.

EXISTING SYSTEMSRetained evidence, versions, events, and recorded outcomes
RECORDARCGoverned lineage, chronology, classifications, and gaps
PROOF OUTCOMESReviewable context and a structured proof package
Vendor-neutral, read-only integrationCustomer-specific deployment reviewAccess and regional controls remain implementation-specificBounded AI; human authority

Platform and Trust Center detail deployment, access, evidence, and regional requirements.

INSIGHTS

Build shared language before evaluating the proof.

Start with the category, then examine how transaction-decision context fragments across banking and payment systems.

Category foundationWhat Is Decision Proof Infrastructure?Define the proof layer, its authority, and its boundaries.First-wedge guideWhat Is Transaction Decision Proof for Banking and Payments?Trace the cross-system context behind recorded transaction decisions.

NEXT REVIEW POINT

See how a fragmented decision becomes replayable proof.

Inspect synthetic evidence, policy, rule or model context, judgment, and approvals without source-system change.

Book Private DemoExplore the Decision Proof LabExplore Platform