Similar to the Contract Audit use case, this firm’s Insurance practice was given the task of analysing and identifying risks within an insurance portfolio that had recently undergone several acquisitions. Many of the policy documents were poor quality, including hand-written notes which meant the other option was to sample documents and accept  that there wouldn’t be full risk coverage.

‍

They invested in fine-tuning a model that could run through the 80,000+ policy documents, extracting 6-20 data points with 99% accuracy depending on the document. They then paid ~£0.02/ page in  processing fees. The output was an API that populated a searchable database their team could then analyse and included a confidence score for manual review.

‍

As a result, the firm could give counsel on 100% of the portfolio and give the client much greater clarity over the risks they were exposed to