Commission statements from multiple carriers are collected, parsed by AI and matched against the policy book automatically — surfacing every missing or mismatched payment instead of burying the team in spreadsheets for days.
Typical results
100%
of statement rows verified
6
carrier formats normalized by AI
1 quarter
for recovered payments to cover the project
Figures reflect typical results for this kind of automation.
The challenge
Every month the brokerage received commission statements from several carriers, each in a different format — PDFs, Excel exports, even scanned tables. Reconciling them against the policy book took a bookkeeper close to three full days, and underpaid commissions regularly slipped through unnoticed.
Before and after
Monthly reconciliation time
−99%Commission rows matched automatically
Under-payments caught
3.3×The automation
We built a pipeline that pulls each carrier statement from email, uses AI extraction to normalize every row regardless of format, and matches it against the expected commissions calculated from the policy book. Matched rows are posted automatically; any gap, rate mismatch or missing policy lands in a discrepancy queue with a one-click dispute email to the carrier.
How it works
- 01
Statements collected
Commission statements from every carrier are pulled from email automatically.
- 02
AI normalization
AI extraction reads each row — PDF, Excel or scanned table — into one common format.
- 03
Match against the policy book
Every row is compared to the expected commission calculated from the policy book.
- 04
Mismatches surfaced
Missing and mismatched payments are flagged with the carrier and policy attached.
- 05
20-minute review
The team reviews a short exception list, and the monthly commission report goes out on time.
Starts from
- Carrier statements
- Policy book
Delivers
- Matched rows
- Discrepancy alerts
- Monthly report
The impact
Reconciliation typically drops from roughly three days of manual work to about twenty minutes of review. Under-payments that used to slip through are caught and disputed every month, and the monthly commission report arrives on the first of the month, automatically.
Questions about this automation
Can it read commission statements in different formats?
Yes. Each carrier sends statements in its own layout, such as PDFs, Excel exports or scanned tables. An AI extraction step reads every row into one common format with policy number, premium, commission rate and amount, so the matching logic works the same way for every carrier.
How does it calculate the expected commission?
The expected amount comes from the policy book: the policy premium multiplied by the agreed commission rate for that carrier and product. Each statement row is compared to that figure, and any gap, rate mismatch or policy missing from the statement is listed as a discrepancy.
What happens to the discrepancies the flow finds?
Each discrepancy lands in a queue with the carrier, the policy and both amounts attached, and the team can send a prepared dispute email to the carrier in one click. Matched rows are posted automatically, and the monthly commission report is built from the reconciled data.