PULSE — Risk.

Score, inspect, act — within the recorded evidence. PULSE exposes a decision explorer, while rationale completeness, model use and downstream actions depend on the producing workflow.

What PULSE owns

• Designed scoring for selected checkout, account, refund, payout and suspicious-pattern events when those integrations supply data
• A rules-first engine; the observed ML indicator is a stub and must not be presented as deployed ML decisioning
• Designed step-up outcomes such as 3DS, re-verification, manual review or hold, subject to downstream integration
• Decision and alert references that can hand work to DOCKET where the relevant case workflow is configured

SENTINEL is the intended governance surface for rules and thresholds, while PULSE consumes applicable versions. Verify that the producing decision actually records and applies the referenced policy version; the architecture alone does not prove runtime parity or prevent drift.

Reading a Decision Record

Compliance Console → Risk → Decision explorer is the observed PULSE surface. Filters and records can expose outcome, score, rule, version, input and correlation fields, but rationale is partial in the reviewed status. Treat absent fields as Not recorded by engine, not PASS. The ML chip is a stub; rules-only decisions should show model version as Not applicable.

A designed governance loop for tuning

Chargebacks and representment

The designed chargeback workflow can link a dispute to decisions, evidence and casework when those records are integrated. A representment pack can include only data actually held, such as 3DS, delivery or communication records. DOCKET SLA, assignment and maker-checker behavior must be verified for this workflow.