Internal tool · PDF parsing, financial automation, document AI
AI-Assisted Commission Processing from Supplier Statements
A sales company working with eight suppliers across France, Switzerland, and Italy had to manually review monthly supplier statements to validate revenues and commission payouts. I built an AI-assisted document analysis tool that turned these recurring PDF statements into structured financial data, replacing a slow manual process with a faster and more reliable workflow for commission validation and financial review.
Challenge
The company received monthly statements from multiple suppliers, often broken down by customer and containing detailed line items for products sold, prices, VAT, discounts, and commissions. These documents were not simply archived — they formed the basis for calculating revenue and validating the company's commission entitlement across a multi-supplier, multi-market business. The process was highly manual. Each supplier used its own document structure, terminology, and layout, which meant the finance team had to inspect every statement individually, identify the relevant figures, interpret the commission context, and then enter the results into Excel. Since payouts differed depending on whether the sales team had placed the order or the customer had ordered directly from the supplier, the task involved not just data entry but applying business rules correctly. That made the workflow slow, repetitive, and vulnerable to inconsistency — and created unnecessary operational dependence on people who had become familiar with each supplier's statement format over time.
Solution
I developed an AI-assisted internal tool to process supplier PDF statements, extract the relevant financial fields, and convert them into consistent structured data for downstream revenue and commission analysis. Instead of relying entirely on manual review, the system turned unstructured supplier documents into usable records that could be checked, compared, and worked with far more efficiently. It was designed around the real operational problem rather than just the document format itself, capturing the information needed to support commission validation across customers, suppliers, and markets. The workflow remained transparent and controllable for the business team. Rather than hiding the logic behind a black box, the tool supported a more reliable review process while removing much of the repetitive document handling and manual spreadsheet entry that had previously consumed time each month.
Technical Approach
The system combined PDF parsing with AI-assisted information extraction to handle supplier statements that varied in structure and layout. Rather than assuming a single standard invoice format, the workflow focused on reliably identifying and normalising the financial fields required for analysis across different suppliers. Key data points such as customer references, products, prices, VAT, discounts, and commission values were extracted and mapped into a consistent schema. This made it possible to compare records across suppliers and support automated calculation workflows based on the company's own commission logic. The output was then prepared for downstream use in Excel and related reporting workflows, allowing the team to keep oversight of the process while significantly reducing the amount of manual checking and entry required. The value of the tool was not only extraction, but structured interpretation: turning messy recurring financial PDFs into data the business could actually use.
Impact
What had previously been a repetitive monthly review process became much faster and more consistent. The tool reduced hours of manual checking and spreadsheet entry, improved reliability across financial records, and made it easier to validate commission payouts across multiple suppliers and customers. It also reduced the fragility of the workflow. Instead of relying heavily on individual familiarity with supplier-specific document formats, the company gained a more systematic and scalable process for handling recurring statements — meaning lower administrative overhead, better consistency, and a clearer path to supporting growth without proportionally increasing manual effort.
Why This Matters
A lot of enterprise AI talk is still obsessed with showy demos and inflated promises. In practice, many of the most valuable applications are far less glamorous and far more useful. This project solved a real operational problem: extracting trustworthy structure from recurring financial documents and applying business logic to it in a way that supported day-to-day work. The value came from speed, consistency, and control — not novelty for its own sake. That is exactly where practical AI earns its place inside organisations. Not by replacing people's judgement, but by removing repetitive document handling, reducing avoidable error, and making important processes easier to scale. This kind of grounded, workflow-level improvement is often what turns AI from an interesting idea into something a business genuinely wants to keep using.