MeridianInsurance

Meridian Insurance underwrites auto and homeowners policies across three regions, processing an average of 2,200 first-notice-of-loss claims a month through phone, web form, and mobile app channels. Every claim, regardless of complexity, passed through a manual intake review before an adjuster could route it to the appropriate team, creating a queue that delayed even straightforward claims.

Intake staff were spending most of their time on data entry and basic severity assessment rather than substantive claims handling, and inconsistent routing meant some claims bounced between teams before landing with the right adjuster. Customers who had just experienced a car accident or home damage were waiting days just to get their claim assigned.

Meridian wanted to automate the intake and triage step specifically, without touching the actual claims adjudication and payout decisions that legally require a licensed adjuster's judgment.

Aivora AI's team spent the first month working with Meridian's claims operations leads across all three regions to document the existing intake criteria and routing rules, then built an extraction pipeline that pulls structured claim data (policy number, incident type, estimated damage, involved parties) from unstructured phone transcripts, web forms, and photos submitted through the mobile app.

A severity scoring model, trained on two years of Meridian's historical claims data, classifies each incoming claim and routes it to the correct regional team automatically. Claims below a defined confidence threshold, or those involving certain high-severity flags, are routed to an intake specialist for manual review rather than being auto-routed, keeping a human decision point wherever ambiguity exists.

The product was built and piloted in one region over eight weeks before expanding to the remaining two regions across the following three months, with Meridian's compliance team reviewing the routing logic at each stage to confirm it met state insurance regulations.