A claim rarely arrives as one tidy document. It is a form, three photographs, a repair estimate and a forwarded email with a paragraph that explains what happened. The adjuster’s first twenty minutes usually go to assembling that into something they can reason about.
One file per claim
Tessera groups everything that arrives on the same thread or reference into a single claim file, classifies each attachment, and extracts a consistent record: policy, date of loss, loss type, parties, amounts and a short narrative. Handwritten forms and phone photos of paper forms are handled; fields written by hand are scored lower and reviewed sooner.
Checks before a person sees it
The agent validates what can be validated mechanically. Is the policy active on the date of loss? Does the claimed amount fit the estimate attached? Are all documents your product line requires present? If something is missing, Tessera sends the request to the claimant and pauses your service-level timer until it arrives.
Judgment stays with the adjuster
The cause narrative is free text, and it is where nuance lives. Tessera extracts it, summarises it in one neutral sentence, and holds it for review when confidence is low, with the source paragraph highlighted next to the summary. It does not decide coverage. Routing is about who looks next, by region, severity and complexity, never about whether a claim is paid.
Keeping a defensible record
Every extracted value links to a region on a source page, and every override records who changed what and why. When a claim is disputed months later, the file reconstructs what the system saw and what the human decided.