The problem
Franchise disclosure documents contain valuable information in a format designed for reading. Turning that material into a usable data source requires extraction, structure, and care about what the source actually says.
My role
As Director of Lending & Risk Products at FranData, I lead product strategy for the lender vertical and manage technology development across the organization. My work connects the needs of enterprise lending clients with the data systems and internal tools that support them.
The document intelligence pipeline is part of that broader work: a multi-stage AI process for converting dense disclosures into structured, machine-readable data.
The approach
The pipeline moves from source documents through extraction and validation into structured outputs. OCR and document processing make the source accessible to downstream stages; structured data makes it usable beyond an individual document.
This work sits alongside research and validation tooling, entity extraction, ingestion automation, and core data platform utilities. The common goal is to make document information easier to work with throughout the organization.
What this demonstrates
The project brings product thinking and engineering together: understanding the information people need, designing the processing steps, and building tooling that connects to the wider data workflow.
This is a public overview of internal work. The implementation and proprietary data are private. Get in touch to discuss the broader approach to document intelligence and data products.