SuryanshChourasia
All work

US commercial insurance broker · 2026

Aegis

I designed and built the broker workspace that extracts ACORD submissions, flags uncertain fields, and prepares each risk for placement.

Frontend developmentInterface design in codeAI integration

A commercial insurance submission could take a broker up to three days to triage before a carrier conversation even began.

Each submission arrived as a long ACORD PDF with business details, revenue, locations, loss runs, vehicles, and property schedules. A broker copied that information into a spreadsheet, checked it, and rewrote it for each carrier. The interface needed to remove the repetition without hiding uncertainty.

Aegis broker platform on a MacBook on a sunlit clay surface, showing active submissions, average quote time, bind rate and pipeline value.
Aegis analytics dashboard showing live KPI counters and portfolio insights for a commercial insurance brokerage.
The operational view. Live counters, carrier hit ratios, pipeline value.
Aegis analytics interface on a desktop monitor, showing premium written, conversion rate, cycle time, loss ratio, and the brokerage conversion funnel.
The portfolio view gives the brokerage one place to compare cycle time, conversion, premium, and carrier performance.

The decisions that made it work

01

Read the form, do not retype it

Field extraction off the raw PDF with a confidence score per field. Anything under the threshold is flagged for a human rather than silently guessed, which is the difference between a tool brokers trust and one they quietly stop using.

02

Make the pipeline visible

Every submission on one board with its stage, its carrier set and its age. The operation could finally see the whole book instead of asking each broker what they were sitting on.

03

Score the risk before the call

Fifty plus factors rolled into a single view, so the conversation with the carrier starts from a position rather than from a blank form.

04

Confidence beats automation

The temptation is to auto-fill everything and look magical. We surfaced the 94% confidence number on every extracted field instead. Brokers are underwriters by instinct. Showing them the uncertainty is what made them trust the rest.

30s

median triage, from 3 days

94%

field extraction confidence

50+

risk factors scored

12pp

bind rate uplift

Built withNext.jsReactClaudePostgresDrizzleServer-Sent Events
Next projectPactGerman hotel group, 47 properties
Want one like this?Landing pages from $999, live in about 10 days.Start a project