AI Email Automation for Insurance Operations

Client: Insurtech startup (confidential)

An internal Outlook add-in with AI parsing that transformed how an insurtech startup handles incoming policy emails — turning a full day of manual data entry into under two hours of review-and-confirm.

The challenge

The operations team received hundreds of policy-related emails daily. Each email required manual reading, classification (storno, update, inquiry), extraction of field values (policy number, client ID, dates), and triggering of the appropriate downstream action — filing the .eml in the document system, initiating a storno operation, updating a record. Processing 80 storno requests alone took a full working day per operator. There was no automation and no tooling — just email, copy-paste, and institutional knowledge.

Our approach

Built an Outlook add-in that intercepts incoming emails and sends them through an AI parsing pipeline. The pipeline classifies the email type, extracts structured field values relevant to that operation type (policy numbers, client IDs, dates, amounts), and pre-fills the appropriate workflow form — storno operation, .eml document filing, record update. The operator sees a sidebar in Outlook with the pre-filled form and confirms or adjusts before submitting. No manual copy-paste; no switching between applications. The AI extraction layer was trained on real operation emails to handle the vocabulary and format variations common in German insurance correspondence.

Tech stack

Results

80% reduction in time spent on email-based operations. The same 80 storno operations that previously took a full working day now complete in 1.5 hours. Extraction accuracy eliminated the copy-paste errors that had caused downstream correction work. The operations team reported that the add-in changed the nature of the role — from data entry to review and decision-making.