AI System for Comparing Payment Applications with Work Records

The problem

Period and cumulative quantities may be confused in day-to-day operations. When payment applications, bill items, approved quantities, site records and previous cumulative totals are scattered across different screens, documents or team records, issues do not become visible in time.

This fragmented data delays preparation of the payment application discrepancy report and requires staff to search for supporting records again by hand. Before making a decision, the supervising engineer and payment application officer have to complete missing fields and verify conflicts one by one.

A common exception is a revised bill item or approved additional work. If an incomplete or incorrect workflow runs without checking this context, valid work may be rejected or a duplicate payment may be made; the output therefore cannot be applied directly.

What we set out to improve

• Number of verified payment application discrepancies
• Rate of outputs corrected or rejected by the supervising engineer and payment application officer
• Time between the source record and the review output

How the system works

1. 01 Define the review scope and the authorised supervising engineer and payment application officer roles
2. 02 Receive the inputs: payment application, bill item, approved quantity measurement, site record and previous cumulative total
3. 03 Check source identity, date, version and data freshness
4. 04 Standardise the bill item and period
5. 05 Compare the quantity measurement and cumulative total
6. 06 Check the exception: there may be a revised bill item or approved additional work
7. 07 Place low-confidence or conflicting records in a separate human review queue
8. 08 Write the source identity, rule, output and time to the audit trail
9. 09 Have the supervising engineer and payment application officer approve, correct, defer or reject the report

Methods we used

• Bill-item arithmetic and document reconciliation
• Source, format, version and required-field checks for the payment application, bill item, approved quantity measurement, site record and previous cumulative total
• A missing-data warning instead of a prediction when inputs are absent or conflicting
• A separate check for the exception outside the main rule
• Joint storage of the source identity, rule result, payment application discrepancy report and human decision

Where people stay involved

Data and security

Access to payment applications, bill items, approved quantity measurements, site records and previous cumulative totals is restricted to the minimum permissions required for each role. Identifiers and sensitive fields are masked where possible, and only the necessary section of a file is sent to the model rather than the complete raw file. Access, output and the reviewers' decision are logged; production use does not begin until the organisation has approved the retention period and data location.

Who this suits

Frequently asked questions