Ingestion Hub, Benchmark's AI document source aggregation service, turning scattered claims and medical paperwork into a single scheduled, secure pipeline.
A leading US-based insurance services company processes claims, medical records, and authorization paperwork for carriers and healthcare networks.
The company needed a reliable pipeline to automatically collect documents from multiple sources and deliver clean, structured data to downstream systems, while maintaining secure integration with third-party platforms.
The company handles high volumes of unstructured PDFs every day, including claims, medical records, and authorizations.
Each document type feeds a different downstream workflow, and delays in collection directly delayed processing and payment cycles.
There was no scheduled pipeline, no audit trail of what was collected from where, and no consistent way to pass extracted data onward.
The company wanted an ingestion backbone it could trust for its AI document processing across insurance, medical, and legal paperwork.
Every operational bottleneck reported was matched to the workstream(s) that resolved it.
Files lived across email, Google Drive, Dropbox, and databases with no single collection point.
Ready connectors for email over IMAP, Google Drive, Dropbox, and databases collect documents from every channel.
Staff pulled files by hand and typed their contents into other systems, introducing errors and delay.
Each document is streamed to an AI extraction service that reads the file and returns structured data, turning unstructured PDFs into usable fields automatically.
Collection depended on people remembering to check sources, so files were missed or processed late.
Automatic sync on a schedule, backed by a BullMQ and Redis queue for dependable asynchronous processing.
Storing logins for many third party sources without proper protection created real exposure.
AES 256 encryption of source credentials, JWT authentication, rate limiting, and full audit logs.
Downstream systems had no reliable way to receive structured data from incoming documents.
An API key guarded external API lets downstream systems pull files and mark them consumed.
Every source flowed into a single, scheduled ingestion backbone with nothing collected by hand.
Downstream teams received clean fields instead of raw documents to re key.
Documents reached downstream workflows within minutes of arriving at any source.
Encrypted credentials, authentication, and audit logs replaced ad hoc storage of logins.
New sources and document types could be added without adding manual effort.
Time from document arrival to availability in downstream systems likely dropped from days to minutes.
Data entry errors from manual re keying were likely reduced significantly through structured AI extraction.
