Loading…
Loading…
Insurance claims triage requires speed, accuracy, and full auditability. Archiet generates a governed claims triage agent: FNOL intake, AI-powered data extraction from claim documents, deterministic DMN severity routing that actuaries control, and a complete audit trail for regulatory examination.
First Notice of Loss (FNOL) triage has three properties that make standard AI agents unsuitable and governed agents ideal. First, the routing decision has real liability: send a £500K commercial property claim to a desk adjuster and you have a coverage dispute; send it to a specialist and you have a handled claim. Second, the decision criteria are explicit: claim type, coverage, estimated loss amount, fraud indicators, and geographic zone determine the routing. Third, the outcome must be defensible to regulators: in the UK, FCA PRIN 6 requires fair customer outcomes; in the EU, IDD requires documented advice processes. Governed agents satisfy all three.
The generated agent implements a standard claims workflow. Intake: FNOL claims arrive via web portal, email, or API from FNMS/FNOL apps. Extraction: the LLM reads the FNOL document, police report, or claimant description and extracts structured fields — claim type, coverage line, estimated loss, date of loss, claimant details, fraud indicators from the document text. Routing: the DMN table maps extracted fields to adjuster tiers (standard desk, catastrophe specialist, SIU for fraud referral, fast-track for low-value). Assignment: the BPMN process assigns the claim to the correct queue, notifies the adjuster, sets SLA timers, and writes the audit record.
Standard fraud detection ML models are opaque — they assign a fraud probability but cannot explain which features drove the score. Regulators and claimant advocates increasingly challenge this. The governed approach: the LLM extracts explicit fraud indicators from the document (inconsistent dates, previously rejected claims, geographic anomalies, unusually precise loss amounts) as structured fields. The DMN policy evaluates these indicators explicitly: if fraud_indicator_count >= 2 → SIU_referral. Every SIU referral cites the specific indicators that triggered it — explainable, auditable, and defensible.
The claims triage agent generates documentation for FCA, PRA, and EU Solvency II regulatory requirements. The BPMN process provides the documented claims handling procedure. The DMN table provides the documented decision criteria. The audit log provides the complete decision trail for every claim. The generated AI governance pack includes the risk classification under the EU AI Act (claims routing systems may qualify as high-risk under Annex III, point 5b — AI in insurance), model card for the extraction LLM, and data processing records required under GDPR Article 30.
Yes. The DMN policy table distinguishes personal lines (motor, home, travel) from commercial lines (property, liability, cargo, professional indemnity) via the coverage_line field extracted from the FNOL. Each coverage line can have different severity thresholds and adjuster routing rules. The table is extensible — add a new coverage line by adding a row to the DMN table, no code change required.
Yes. The BPMN process includes an override task that claims managers can invoke from the operator console. Overrides are logged with the manager's identity, the original routing decision, the override decision, and the reason. Override rates are tracked in the analytics dashboard — a high override rate for a specific rule signals that the DMN table needs adjustment.
The BPMN process includes a catastrophe mode flag that routes all claims from a declared CAT event to a specialist queue regardless of individual claim characteristics. The CAT flag is set via the operator console and takes effect immediately. The Celery worker queue for claim processing scales horizontally — the generated docker-compose.yml includes a claim_worker service with configurable replica count for CAT surge processing.
Most AI invoice automation is a black box — the model decides, you can't explain why. Archiet generates a governed invoice approval agent: the approval flow is BPMN (auditable), the decision logic is a DMN policy table your finance team can edit without code changes, and the AI only reads documents — it never decides the outcome.
Enterprise procurement teams need approval automation that can be explained to auditors, edited by business analysts, and trusted in regulated industries. Archiet generates a governed procurement agent: BPMN workflow, DMN policy table, bounded LLM for document reading, and a complete audit trail from day one.
Loan approval decisions need to be explainable to regulators and customers. Archiet generates a governed loan adjudication agent: the BPMN workflow defines the process, the DMN credit policy table defines the approval criteria, and the AI only reads documents — it never decides the outcome.
free plan. No credit card required. Generate your first compliant architecture blueprint in under 10 minutes.