The starting point
Underwriters were spending the majority of their day on submissions that were never going to bind - out-of-appetite risks, incomplete submissions, broker mismatches. The MGA's hit rate was healthy but the throughput was capped by attention, not capacity.
What we deployed
A decision intelligence layer that scores every inbound submission against the MGA's appetite, completeness, and historical bind probability. Bind-ready submissions surface to underwriters first, with a pre-built risk pack. Out-of-appetite submissions route back to brokers with a reasoned response. Edge cases route to senior underwriters with the model's reasoning attached.
Outcomes
Bind-ready throughput per underwriter doubled. Time spent on out-of-appetite risks fell 38%. The targeted portfolio's loss ratio improved 11 points within two renewal cycles, driven by sharper risk selection at the front of the funnel.