Extracting and structuring information from PDF documents

Founding engineering team and F-CTO for a startup

CHALLENGE

Extract text from insurance coverage documents (EOC) available only as PDFs, with varying formats and sources, making it understandable only through visual layout. Converting this data into a structured format was essential for multiple applications, including improving accessibility for the public.

SOLUTION

Our AI-based solution included a powerful PDF parser that seamlessly adapted to diverse document types and correlated text using visual cues. Then, we leveraged cutting-edge machine learning (ML) functionality to restructure and normalize the raw text for specific benefits.

Additionally, we developed tools to standardize benefit names across all documents, enabling the delivery of structured information through a user-friendly API.

RESULT

Hundreds of documents ingested, covering various types of documents, totalling hundreds of benefits. Key information automatically extracted and easy to integrate into various applications. Core functionalities currently in use in five different products.