ParaLegal

Inspiration

Lack of monetary resources to consult a lawyer.

What it does

AI tool to get primary law and secondary data with references to old judgments to supplement strong cases for all our legal battles at a cheaper rate.

How we built it

Scraped millions of old history cases. Kindo.ai for custom GPT. Fed that into LLMs. LegalBert, PineCone for Embeddings. Streamlit.io for the front end. LangChain. Llama-Index for indexing.

Challenges we ran into

Accuracy of model. Varied datasets. Custom embeddings. End-user input was required.

Accomplishments that we're proud of

Data gathering for entire US law and case studies using simple Python scripts.

What we learned

Kindo.ai platform's custom GPT

What's next for ParaLegal

Battle of LLMs between Plaintiff and Defendent. Understanding Contracts. Having it published for multiple languages across the world.

Built With

• firestore
• kindo.ai
• langchain
• legalbert
• llama-index
• pinecone
• python
• streamlit.io