Legy.md
AI legal assistant for Moldovan legislation
Tech stack
- Next.js
- TypeScript
- RAG
- OpenAI
- PostgreSQL
- Vector DB
A production AI assistant providing instant, contextual answers about the Civil Code, Criminal Code, Labor Code and the full body of Moldovan law. Built from the ground up — RAG architecture, curated legal corpus, conversational UX.
Gallery
Highlights
- Custom RAG pipeline over Moldovan legal corpus
- Multilingual (RO/RU) legal Q&A
- Production deployment with usage tracking
- Citation-grounded answers with source linking
My case study
I built Legy.md as a serious legal research product, not a chatbot demo. The goal was to make Moldovan legislation searchable through a conversational interface while keeping answers grounded, multilingual, and easy to verify.
- Retrieval pipeline over Moldovan legislation
- Legal RAG
- Multilingual legal Q&A experience
- RO / RU
- Production product with usage tracking
- Live
The challenge
Legal information is dense, fragmented, and hard to navigate for non-specialists. A simple AI wrapper would create risk, so the product needed a structured corpus, careful retrieval, and source-aware answers.
What I built
I designed the RAG flow end-to-end: legal document preparation, embeddings, retrieval, answer composition, multilingual UX, deployment, and operational monitoring.
Business outcome
Users can ask practical legal questions and receive contextual answers backed by relevant sources, making the product feel useful, credible, and commercially defensible.
Related services