Context
An insurer wanted to identify the best OCR and semantic analysis solution to automatically compare PDF documents of insurance contracts from multiple suppliers.
Challenge
- Difficulty extracting and structuring key contract data in PDF format.
- Need to benchmark several LLMs (Claude, Mistral, Mistral, ChatGPT, NoNote, Dolphin) on criteria of precision, clarity and comprehensiveness.
- Need for a simple interface to visualize the results, interact with the contents and refine the analysis.
StratBoard Solution
- Deployment of a custom platform to load PDFs, view them and compare them.
- Integration of several LLMs to execute the same queries on documents and assess the quality of responses.
- System for ranking performance by business criteria (extraction of clauses, calculations, obligations, exclusions...).
- Document chat functionality to refine analysis or ask specific questions.
Results
✅ Structured and contextualized multi-LLM comparison.
✅ Intuitive interface for viewing contracts and their analyses.
✅ Decision support on choosing the best LLM for each use case (precision vs cost vs speed).
Conclusion
StratBoard facilitates the selection of the best LLM for documentary analysis in insurance, by combining benchmark, business visualization and intelligent interaction with content.
