Inspiration

Reading long SEC documents sucks! Especially all the specialized jargon.

What it does

This reader allows you to look up specialized words in the context of the document.

How we built it

Llama-index + replit on the backend. NextJS + Vercel for the frontend

Challenges we ran into

PDF parsing is enormously difficult. I built a series of custom parsers on top of Unstructured that all failed to catch the 5% of edge cases for terms defined in a document. Differing more of the parsing to LLMs helped a lot.

Accomplishments that we're proud of

Nice UI. Prompt engineering that delivers. A working app!

What we learned

PDF data ingestion is no joke!

What's next for LazyLegal

Learning more about the fragmented legal tech space

Built With

Share this project:

Updates