Efficient next-level document processing in banking
Context & Objectives
A leading bank in Belgium was looking for a way to increase the efficiency of its back-office process by automatically dispatching incoming documents to the correct client folder. Their current manual approach no longer fit their needs due to the limited number of records that the team could treat in place.
The goal was to ensure that all incoming documents would be treated within a day while ensuring that we could link each to the right client folder and assign each to the correct processing team afterward.
Approach
We started by defining the project's scope and its complexity (number of documents to treat, characteristics of the categories, information to extract per document). We used Natural Language Processing (NLP) techniques to automatically classify any incoming file into the right category.
We then implemented various text extraction methods to extract the right information per document and link it to the correct client folder.
Finally, we packaged the solution so that all documents could flow with a single call through an API from the necessary department.
Results
In less than three months, we managed to deliver the following results:
Considerable time savings: our tool drastically diminished the processing time needed to treat incoming documents no matter their size or content and allowed the client team to handle all incoming documents daily
Improved classification: moving to a model-based procedure helped to increase the precision of the classification that was made based on the document’s content
Significant cost reductions: with a document dispatching process now fully automated, our client could save up about 3 FTE’s per year
