How Ethias automated 70% of incoming claims documents
Health-claim reimbursement typically starts with paperwork in the form of medical invoices, certificates, etc. For Ethias, that means processing around 120,000 documents every year before the relevant data can reach its claims systems. The goal was to remove as much manual document handling as possible without giving up control over exceptions.
120,000
documents per year
70%
fully automated
Challenge
Incoming documents in claim are rarely clean, single-document PDFs. Different document types can be bundled together, layouts vary, and important information can sit inside complex tables. Teams first had to identify the right documents and then capture the data needed for reimbursement, creating a high-volume manual bottleneck.
Approach
Skwiz became the intelligent document-processing layer in the claims flow. It separates incoming files into individual documents, classifies the relevant document types, extracts the required fields and tables, and sends uncertain cases to a claims handler for human review.
Together with Sagacify, the workflow was built step by step around the claims process. The initial scope focused on four document types, with splitting, classification and detailed extraction combined with validation rules. The setup was designed so additional document types can be easily added in the future.
Result
The live workflow processes about 120,000 documents per year. Around 70% are handled fully automatically, reducing the workload substantially while keeping manual validation available for the remaining exceptions.