Automated submission triage
Using models to read the files brokers send, extract their data and rank them by appetite and likelihood of winning, so the underwriter spends time on the business that deserves it.
A commercial lines underwriter often receives many more submissions than it can seriously study, and a large share of them do not match its appetite or have no chance of success. Automated triage addresses this bottleneck. Models read emails and attachments, identify the insured, its business, the amounts, requested covers and loss history, match these data with external sources, then give each submission a score combining fit with appetite, apparent risk quality and likelihood of binding. Out-of-appetite submissions are declined quickly, the most promising move to the top of the queue. The gain is not just time saved: speed of response is a decisive competitive advantage with brokers, and a submission handled in a day is more likely to be won than one handled in two weeks. The risk is that of any automatic filter: a model trained on past decisions reproduces their biases and may systematically discard unusual but profitable risks. The underwriter must therefore monitor what the triage rejects, not only what it proposes.
Textbook case: a team receives 1,200 submissions a month for a study capacity of 300. Automated triage declines 500 out-of-appetite files up front, ranks the rest, and the team answers on average in two days instead of nine; its hit ratio on the business studied rises from 12 to 18 percent.
Follow the research
One letter a week. First, the week's underwriting note, when one is published: a dated, argued decision, signed by Paul Bernardski. Then what the observatory published on cyber risk, political risk and insurance. No account, no tracking.
No spam, one-click unsubscribe, no data ever sold.