Automatically extracting values from financial tables and notes and mapping to custom taxonomy
A large American financial services & rating agency had to Process ~200 thousand annual reports of banks
Around 2 million pages in 10+ languages and in multiple reporting formats & styles
High variations in the descriptions and line items that need to be mapped to a single taxonomy
Data normalization and transformations are needed post extraction, 100+ trained analysts performing manual tagging
Quality is critical as it powers some important financial models
Leveraging the pre-trained table understanding and financial statement understanding models to deliver the first version of the solution within days.
Trained the client’s SMEs on Botminds AI platform. SMEs used point & click interfaces from day 1 to improve the accuracies.
Upstream and downstream custom integrations made to make Botminds AI to work seamlessly with client's current workflow
See Botminds Unified platform in action