Case Study: a financial data and market intelligence provider links unstructured company presentations to its GICS database with Flexi IT's AI-powered entity resolution pipeline

a financial data and market intelligence provider achieves 85% match rate with Flexi IT

Flexi IT developed an AI-powered entity resolution pipeline for a financial data and market intelligence provider. The client had two disconnected datasets: a large archive of unstructured company presentation PDFs and a structured company database. Their challenge was to reliably link each document to a corresponding company record despite having no shared identifier, inconsistent company name formats, and many scanned or image-based PDFs.

The solution built by Flexi IT was a multi-stage pipeline featuring resilient PDF text extraction, cost-aware sampling, and multi-metric matching. It used a combination of fast local parsers and AI models to extract company data, then matched it to the database using a cascade of string similarity methods. The system achieved the target match rate of at least 85%, made the entire archive processable, and ensured processing costs scaled predictably with document count, not page count, transforming the static files into a searchable intelligence layer.