Bordereaux processing

Challenge  

• Parsing Bordereaux documents, critical for data extraction, is challenging due to lack of standardization, multi-sheet structures, and inconsistent headers
• Existing tools struggled to automate this complex task efficiently and accurately
• Complexity made processing workflows extremely manual and prone-to-error

Solution  

• Developed a state-of-the-art, three-pronged asynchronous parsing approach using AIP and GenAI (deterministic logic, code-generation, full GenAI)
• Implemented a multi-model consensus "chain of thought" mechanism to evaluate the best strategy for each file
• Utilized TypeScript Functions, Logic Files, and custom pipelines, pushing the boundaries of Foundry's AIP capabilities

Results  

• Achieved an unprecedented 78% parsing accuracy, outperforming market-leading tools by two-fold
• Accelerated manual parsing workflows by 10x, processing complex multi-sheet Excel files in under 2 minutes
• Demonstrated the transformative potential of combining AIP and GenAI within Foundry