Risk scoring, anomaly detection and Ask Vetra.
Vetra ships AI into the daily workflow, not as a side panel - every claim is risk-scored on submission, every flag is audit-grade, and Ask Vetra sits one keystroke away.
Per-claim risk scoring
Every claim receives a risk score on submission. Triggers include:
- Duplicate billing across providers or dates.
- Impossible service dates (e.g. dates of service before plan effective date).
- Upcoding patterns (procedure codes that consistently sit above the norm).
- Rapid-fire submissions from the same provider for the same enrollee.
- Cross-tenant patterns picked up from the platform-wide model.
The risk panel on a claim - score, contributing rules and historical context
Screenshot - coming soon
AiRiskEvent - the audit primitive
Every flag is persisted as an AiRiskEvent: the rule, the inputs, the score and the time. That gives compliance a defensible trail and powers the Fraud & anomaly report.
| rule | The trigger that fired (e.g. DUPLICATE_LINE, INELIGIBLE_DATE). |
|---|---|
| inputs | The claim, enrollee and provider IDs the rule examined. |
| score | A 0-100 risk contribution from this rule. |
| createdAt | When the rule fired. |
Fraud & anomaly report
- Trends per provider, procedure and enrollee.
- Top contributing rules across the period.
- Drill-through to the original claims and AI events.
The Fraud & anomaly report - provider league table, top triggering rules and trend chart
Screenshot - coming soon
Ask Vetra - the in-app copilot
Ask Vetra is a draggable widget in the lower-right of the HMO and provider portals. It can:
- Summarise a claim, encounter or PA.
- Surface similar past cases and decisions.
- Draft a query response or a rejection rationale.
- Translate clinical notes into plain language for a finance reviewer.
The Ask Vetra widget open on a claim - summarising the encounter and suggesting a query
Screenshot - coming soon
Ask Vetra in action
Summarise a claim, find similar past decisions, draft a query reply.
Walkthrough - coming soon