How Solidaris Assurances eliminated a 3-month indexing backlog
Solidaris Assurances, the insurance company of the Solidaris group, receives a continuous flow of documents that must be indexed before files can be handled. With manual indexing, the team was running around 3 months behind, and beneficiaries felt that delay directly. The goal was to process documents as they arrive, without losing control over exceptions.
3 months
backlog eliminated
98%
extraction accuracy
Challenge
Around 180,000 documents arrive every year, often bundled in large files that mix many document types. Indexing meant finding the relevant documents in that flow, reading detailed information that often sits in complex tables, and looking up the matching client in the internal database. Done manually, this work couldn't keep pace with incoming volumes, and the backlog kept growing.
Approach
Skwiz splits incoming files (large PDFs) into individual documents and classifies them to identify the 20 document types in scope. It extracts the required fields, including complex tables, and matches the extracted client details against the Solidaris Assurances client database, so each document is linked to the right client automatically.
Documents that are not approved automatically go to a validation interface, where the team can review and correct each step: splitting, classification, extraction and client matching.
Craftzing managed the project and built the workflow with Skwiz around Solidaris Assurances' existing processes and systems.
Result
The backlog is gone. Documents are now processed almost as soon as they arrive, so beneficiaries no longer wait months to be reimbursed. Extraction reaches 98% accuracy, and documents that need a check stay under human control.