FNOL AI Hub for Multi-Line Insurance (Motor, Property & Travel)

The insurer operated in multiple countries with motor, property and travel lines of business. On paper, FNOL was “digital” — customers could email, call, submit forms or use partners’ portals. In reality, each line had its own intake quirks and manual workarounds.

• Motor FNOLs landed as long free-text emails with photos and policy numbers buried inside.
• Property FNOLs arrived through brokers and TPAs with mixed-quality spreadsheets and PDFs.
• Travel claims were captured by a third-party platform and sent nightly as unstructured files.
• Every team keyed the same data into the core claims engine in their own way.
• Supervisors had no single view of “today’s new losses” by segment, channel or severity.

The leadership team wanted a unified FNOL brain that would sit in front of the core system, standardize intake and make AI useful without rewriting all the back-end platforms.

Define and implement an FNOL AI hub that would:

• Accept FNOL from any channel — email, app, web, broker feed, call summaries.
• Use AI to extract policy, incident and coverage-critical details in minutes.
• Standardize the FNOL data model across motor, property and travel lines.
• Auto-triage severity and route to the right claims queues and partners.
• Feed clean, consistent FNOL records into the core policy / claims systems.
• Provide a single dashboard for supervisors to see new losses and risk signals.

Before Xovity, FNOL looked different in every corner of the business. Each line had “its own” way of capturing the same basic facts: who, what, where, when and how bad.

• Motor claims handlers retyped email content into the core claims UI while calling the customer back.
• Property teams printed broker spreadsheets and manually flagged incomplete files with sticky notes.
• Travel FNOLs arrived as overnight batches; high-severity cases were sometimes noticed hours later.
• Fraud teams had no early view of suspicious patterns at the intake stage.
• Reporting teams reconciled three different FNOL “truths” before each monthly steering committee.

Xovity introduced an FNOL AI hub — a centralized intake brain that sits in front of the existing policy and claims engines. Instead of each line running its own process, all FNOL journeys now pass through a common AI and workflow layer.

The same pattern is now being cloned for new countries and lines of business by simply changing configuration and rules — not rewriting the architecture.

Phase 1 — Discovery & Pattern Definition

• Walked through live FNOL calls and email queues with motor, property and travel teams.
• Classified intake types: direct email, portal uploads, broker files, third-party feeds.
• Defined a standard FNOL skeleton (policy, incident, coverage, risk) shared across lines.

Phase 2 — FNOL Data Model & AI Prompts

• Designed Dataverse tables for FNOL header, participants, locations, vehicles, assets and travel segments.
• Built Azure OpenAI prompt templates that “read” free text and return structured FNOL JSON.
• Created test sets with real historical FNOL data (anonymized) to calibrate extraction quality.

Phase 3 — Workbench for Claims Handlers

• Developed a model-driven app where handlers can see raw content, AI suggestions and the structured FNOL side-by-side.
• Added guardrails: mandatory human confirmation for certain fields (e.g., liability, coverage position).
• Implemented quick actions: “create claim”, “send clarification email”, “route to partner”.

Phase 4 — Core System Integration

• Integrated with the motor, property and travel core platforms via secure APIs and staging tables.
• Mapped standard FNOL schema to each platform’s required code sets and data structures.
• Ensured idempotent claim creation to avoid duplicates when emails are re-sent or re-processed.

Phase 5 — Rollout, Training & Continuous Tuning

• Started with one motor country and one property country; then extended to travel line.
• Set up a monthly “FNOL AI Council” to review extraction errors and tune prompts and rules.
• Established dashboards and operational KPIs to track adoption and triage performance.

Within the first six months, the FNOL AI hub became the front door for all new losses, independent of the back-end system in use.

• Pattern, not just a project: The FNOL AI hub is a reusable pattern — the client now uses it for new geographies and lines without re-architecture.
• AI + Rules + Humans-in-the-Loop: We deliberately combined AI extraction, classic rules and handler oversight to keep decisions auditable.
• Multi-Line by Design: Data model and routing logic were built for motor, property and travel from day one.
• Core-System Friendly: No rip-and-replace — the hub creates cleaner FNOL records in the systems the client already runs.
• Governed & Observable: Full telemetry on AI suggestions vs human overrides, feeding continuous improvements in both process and prompts.

Xovity’s role was not to “sell AI”, but to shape FNOL into a predictable, multi-line pattern that the insurer can now roll out and govern at scale.