Change or challengeModel · rule · policy · mapping
DECISION PROOF INFRASTRUCTURE
When a critical rule, model or policy changes, know which decisions are affected — and what can still be proven.
RecordArc connects the retained evidence around past decisions so an organisation can see what was affected, what happened, what is missing or conflicting, and what needs human review — without changing the systems that made the original decisions.
Continuous Decision Challenge ReadinessRead-only by designVendor-neutralNo writeback
FROM CHANGE TO REVIEWABLE PROOF
01 02 Affected decisionsDirect · potential · unaffected03 Proof stateAnswered · blocked · conflicting
THE PROBLEM IN PLAIN LANGUAGE
Something changed after decisions were made. Which past decisions were affected, and can you still explain why?
The evidence may still exist, but it can be distributed across systems, tied to different versions, incomplete, or contradictory. That makes a later review difficult even when the original records were retained.
Transaction systemsAuthentication systemsRules and modelsWorkflow and case systemsConfiguration and version recordsHuman review recordsDownstream outcome systems
Identify the evidence-linked population connected to the challenged model, rule, policy or mapping through retained lineage.
Separate answerable proof questions from partial, blocked or conflicting ones.
Expose missing versions, approvals, source roles or chronology instead of hiding them.
Surface cross-system divergence while preserving the authority of each retained source.
Create a human-review cohort without changing, closing or remediating source decisions.
HOW RECORDARC WORKS
Follow one clear path from a challenged control to reviewable proof.
Start with the ordinary-language task. The RecordArc capability name and deeper evidence remain available underneath it.
- 01
What changed?
Start with the challengeA model, rule, policy, configuration or source mapping changes or is challenged after decisions have already been made.
Risk Model 2.3 is identified for review. - 02
Find affected decisions
Decision Impact RadarFind the decisions directly affected, potentially affected or supported as unaffected by the retained evidence.
Separate the four synthetic payment decisions by retained impact evidence. - 03
Understand one decision
Decision Proof TwinSee the retained evidence, versions, human actions and outcomes around one decision.
Open PAY-24092 and follow its retained payment story. - 04
Test what can be proven
Challenge ReadinessSee which questions are answered, partial, blocked, conflicting or not applicable — and why.
The decision is explainable in part, but its reviewer identity is missing. - 05
Find system disagreements
Cross-System DivergenceExpose contradictory retained outcomes or evidence without deciding which source is correct.
PAY-24092 retains APPROVED in one source and DECLINED downstream. - 06
Review and carry the proof
Human Review Cohort · Portable Proof CapsuleIdentify the decisions that need authorized investigation and provide a bounded evidence snapshot with limitations and integrity information.
Place three decisions in human review and verify the PAY-24092 proof snapshot.
TRY A REALISTIC SYNTHETIC CHALLENGE
Risk Model 2.3 may require review. Find the payment decisions that could be affected.
A bank discovers that Risk Model 2.3 may require review after payment decisions have already been made. Find which past decisions may be affected, understand one decision, test what the retained evidence can prove, expose disagreements, and identify what needs human review.
Needs authorized human review
risk-model/2.3Needs authorized human review
risk-model/2.3Needs authorized human review
version lineage incompleteSupported outside the review cohort
risk-model/3.0RecordArc found 2 directly affected decisions, 1 that may be affected because exact model-version lineage is incomplete, and 1 with retained evidence supporting exclusion. 3 decisions need authorized human review.Open the interactive investigation →
FIRST IMPLEMENTATION WEDGE
Banking and Payment Decision Assurance.
The first public product experience focuses on payment decisions because model, rule, authorization, risk, configuration and human-exception evidence naturally span multiple systems and later review teams.
Payment and transaction event contextAuthentication and authorization recordsRisk, rule, model and configuration versionsHuman referral, override and recorded outcome
Review Banking & Payment Decision Assurance or inspect the separate synthetic AML applicability example.
NEXT REVIEW POINT
Test the product thesis yourself.
Use the frontend-only synthetic Lab to investigate a model, rule or mapping challenge, inspect a Decision Proof Twin and verify a proof capsule before deciding whether the problem is worth solving in your environment.