Finance reconciliation: from dozens of Excel files to 1 auditable report

The problem

Every month, before supplier invoices are paid, the finance team compares them against the company's internal data. Until now, 2 specialists built a register by hand across around 60 separate files, 1 per supplier, mostly by copying data out of the billing system. That took around 8 working days per cycle. The process was slow, manual copying carried a high risk of error, and reconciling the data made auditing difficult.

The solution

Pinpoint AI replaced around 60 separate files with 1 consolidated report. 1 billing data extract is processed by a deterministic system, which calculates each supplier's commission fee according to the contract terms, runs built-in checks, and records anomalies on a separate sheet. The solution also checks itself by comparing the result against a reference month. Only if those checks pass does the process continue. A specialist gives the final approval.

How it works

1. Instead of around 60 separate files, the system processes 1 consolidated billing extract.
2. The AI calculates each supplier's commission fee according to the contract terms, whether that is a percentage or tiered.
3. Built-in checks and control totals confirm the sums agree. The system checks itself each period and continues only if those checks pass.
4. Anomalies are recorded on a separate sheet: for example, an invoice from a supplier the company no longer works with, or a new client not yet in the register. The anomaly list grows with each cycle, so the next check is more accurate.
5. The system compares the result against the previous reference month and reports only the differences rather than correcting them itself.
6. The process logic and documentation live in a single knowledge base, so changes to the business logic take effect automatically.

Results

• 1 report instead of dozens of files: 1 consolidated report replaces around 60 separate Excel files.
• Around 8 working days of manual work removed each cycle: the register building that previously ran across dozens of files is eliminated.
• Built-in checks and self-verification: the AI compares the result against a reference month and continues only when the totals agree, preventing the errors that come from copying data by hand.
• Anomaly detection that compounds: the anomaly list grows with every cycle, making the next check more accurate.
• The finance team keeps control: the AI reports differences against the reference report rather than correcting them itself.