Regbase: Al-native Legal Research

Could you build this?

Partial A web interface wrapping search and LLM synthesis across legislative documents is vibe-codeable, but scraping, parsing, and normalizing proposed bills, regulations, and grant notices across hundreds of local, state, federal, and global government portals is a major data ingestion hurdle. Sustaining reliable parsers across idiosyncratic government websites requires dedicated data engineering.

What it would actually take: The stack needs a fleet of distributed scrapers targeting federal (e.g., Congress.gov, Federal Register), state, and municipal legislative databases, converting PDFs and disparate markup into normalized legal schemas. An indexing pipeline using hybrid search (BM25 + dense legal embeddings like LegalBERT) feeds a RAG pipeline designed to track amendment diffs and legislative progression. The hard part is maintaining hundreds of brittle public-sector scraping endpoints and verifying statutory cross-references, requiring dedicated web scraping infrastructure and legal taxonomy expertise.