AI SIU Referral Agent

How Cassidy automates SIU Referrals using AI

Step 1: Trigger on new claim or FNOL event

The Workflow activates when a new claim enters your system—whether from a portal submission, contact center intake, or TPA feed—pulling in policy details, coverage status, and loss information.

Step 2: Enrich with internal and external data

Cassidy pulls claims history, prior losses, SIU records, and policy data from your systems, then enriches with consortium sources like ISO ClaimSearch, public records, and provider databases to build a complete picture.

Step 3: Apply governed fraud rubrics

The Agent evaluates the claim against your codified red-flag catalog—LOB-specific indicators, regulatory definitions, and watchlists—firing rules that produce reason codes and link directly to supporting evidence.

Step 4: Generate risk score with reason codes

Cassidy scores the claim for fraud propensity using ML models, returning a risk score alongside the top drivers so your team sees exactly what triggered the flag.

Step 5: Triage and route by score band

Claims are segmented into work queues based on score thresholds: low-risk claims fast-track to processing, amber claims route to adjuster review, and high-risk claims escalate toward SIU referral.

Step 6: Generate SIU referral package

When “reasonable belief” thresholds are met, Cassidy auto-generates a referral package with synopsis, fraud type, party details, amounts, reason codes, and evidence attachments—formatted to meet state regulatory requirements.

Step 7: Human-in-the-Loop review and routing

Adjusters or SIU leads review the referral, with the ability to approve, suppress, or escalate—each action logged with documented justification that feeds back into model improvement.

Step 8: Log audit trail and sync to case management

Cassidy writes the complete decision trail—data sources, triggers fired, scores, human actions, and timestamps—to your case management system, ready for regulatory examination or legal review.