From data deluge to debt recovery triumph: Indium's AI solution boosts efficiency 16x with 95% accuracy (Leveraging and LLM)

Project Overview

The client wanted to extract, classify, and summarize value interest fields from more than 50 PDF (legal, finance, and due diligence) documents. The client used AI in combination with other plug-and-play text extraction solutions from Amazon and Azure, which did not meet their requirements.

About Client

The client is a leading international debt recovery organization, that collaborates with 117 law firms across 100+ countries. Their customer-centric approach prioritizes out-of-court solutions for debt recovery, fostering strong business relations. Backed by skilled in-house lawyers, they’ve served 3,000+ European clients, managed 10,000+ positions, and 73 million debts in 2 years.

Business Challenges

  • The client encountered difficulties extracting critical information from a variety of unstructured legal, financial, operational, due diligence, and performance-related documents.
  • These documents are collected from diverse locations, and they frequently come in different languages, formats, structures, and currencies.
  • The extraction process required adapting off-the-shelf tools such as Textract, Cognizer, and Document AI to meet their requirements.
  • However, because these text extraction systems are limited to single-page data extraction, they experienced difficulties when dealing with legal, financial, or due diligence documents.
  • Furthermore, managing the large number of data fields inside their documents adds to the complexity.