Designed an AI-assisted underwriting workbench for a home insurance business line estimated at $26M–$148M in annual revenue.
The tool unified risk, property, policy, and portfolio context into a flexible decisioning workspace — replacing scattered tabs, spreadsheets, and systems with a modular interface underwriters could customize around how they actually work.
I shaped the information architecture, workflows, UI patterns, review states, exception handling, and decisioning surfaces. The system combined machine learning, traditional data processing, and scoped AI querying to surface the right context at the right moment — augmenting judgment while keeping high-accuracy data structured, traceable, and reviewable.
Impact: reduced average decision time from 25 minutes to 10 minutes, improved decision quality, and helped flag unseen risks that could prevent costly underwriting mistakes.
