Best Practices for Extracting Data from Court Filings with AI?

[Philosophia]

I’m working on a project for a law firm to automate the extraction of relevant details from lengthy and disorganized monthly court filings. Initial attempts with models like ChatGPT haven’t delivered the needed accuracy and often miss critical details.

Given this challenge, I’m unsure whether fine-tuning a model is required here or if adopting a RAG approach would be effective enough. Does anyone have experience with similar data extraction tasks? What techniques or models would you recommend to improve accuracy and reliability in this context? Any insights or suggestions would be greatly appreciated. Thanks!