I'm a software developer in India exploring a legal tech startup idea called LawSetu. The Indian legal system is overwhelmed—50M+ pending cases, a 3M lawyer shortage, and skyrocketing AI adoption (79% of pros now use it, up from 19% last year). Law firms and solo practitioners waste hours on research, case management, and drafting, often juggling fragmented tools.
LawSetu is an AI platform inspired by the Chatlaw paper (arXiv:2306.16092), using a multi-agent system to help lawyers manage cases. Key features:
Persistent Case Chats: Each case is a dedicated "chat file" with its own memory—short-term convo history + long-term facts/docs stored in PostgreSQL/Qdrant.
Agent Workflow: Central orchestrator coordinates agents like LegalAssistant (builds knowledge graphs), LegalResearcher (RAG-based vector DB searches), SeniorLawyer (reasoning/analysis), and LegalEditor (drafts reports).
Tech Stack: Python/FastAPI backend, Qdrant for vector search, focused on Indian law (e.g., IPC, CrPC, case precedents).
MVP Focus: Starting with intake + intelligent search to prove value for mid-size firms.
Target users: Lawyers in India handling corporate, litigation, or compliance work. Pricing could be subscription-based (~₹500-2000/month per user), with freemium for solos.
YC has backed awesome legal tech like Ironclad and Casetext—curious if this fits the mold. Is this something you'd use (or recommend to lawyer friends)? What's missing? Any pitfalls in legal AI for emerging markets? Early feedback on market fit or tech would be gold—I've got a rough MVP sketch and am validating before full build.