Challenge
One of Denmark’s largest property management platforms wanted to leverage AI to help users classify expenses correctly. The challenge was the sheer complexity of the chart of accounts: with several hundred accounts to choose from, users frequently misclassified purchases. This created significant issues for budgeting and financial planning, as incorrect postings made it difficult to accurately understand spending across activities from year to year.
The Problem
- Millions of invoices processed manually every month
- 30% of invoices are posted to the wrong accounts
- Two full-time employees dedicated to manually handling and correcting postings
- Poor budgeting accuracy due to frequent misclassifications
- Inconsistent and low-quality bookkeeping descriptions
Solution
We developed a comprehensive AI-powered invoice processing system (Faktura håndtering):
- Intelligent Account Selection: The AI agent interprets user input and selects the appropriate account, even when the text contains spelling mistakes, missing information, or inconsistent phrasing.
- Automated Posting Description Generation: The system produces clear, consistent, and standardized posting descriptions based on user intent. This ensures that all entries follow the same format and taxonomy, significantly improving bookkeeping quality and readability
- Intelligent OCR Processing: The solution reads invoice data using advanced OCR and ensures that all relevant information is accurately extracted and applied in the accounting process. Nothing is missed, and data quality is consistently high.
