Case Study: credX accelerates real estate transaction analysis with deepset's Haystack Enterprise Platform

credX reduces document screening time by 80% with deepset

credX, a digital commercial real estate lending platform, faced a significant challenge with its document-heavy analysis process. For each transaction, analysts spent days manually reviewing over 50 documents totaling more than 1,000 pages to extract hundreds of data fields. They partnered with vendor deepset to build a custom Gen AI solution that could deliver the required near 100% accuracy, which traditional intelligent document processing tools could not meet.

Using the Haystack framework, deepset helped credX build a multi-stage AI assistant that processes complex real estate documentation. The solution extracts over 200 data fields with 99.5% correctness, reducing document screening time by approximately 80% and saving analysts 40% of their hours. This not only improved efficiency but also created a new SaaS revenue stream for credX, which now licenses the tool to other commercial real estate players.

credX