Skip to content
Back to work

Live in production

Deedly — AI Legal Document Platform

Contributing developer on an AI-powered legal document management and contract analysis platform.

  • Next.js 14
  • TypeScript
  • Supabase
  • BullMQ
  • Gemini AI
  • BlockNote
  • Tailwind CSS
Visit live site (opens in a new tab)

The problem

What needed solving

Reading and understanding legal contracts is slow, expensive, and inaccessible. Individuals and businesses sign agreements without fully understanding the risks, obligations, or missing protections.

Constraints

What the build had to respect

  • AI analysis pipeline must handle large documents without blocking the UI
  • Multi-provider AI routing to avoid single-vendor lock-in and manage costs
  • Real-time progress tracking for long-running analysis jobs

Architecture

How it fits together

frontend

Next.js 14 App Router with BlockNote rich-text editor, streaming AI chat, and real-time analysis progress

backend

Next.js API Routes with BullMQ background job queue for the 6-stage analysis pipeline

database

Supabase (PostgreSQL + Row Level Security) for documents, analysis results, and user data; Supabase Storage for file uploads

deployment

Vercel with Redis-backed BullMQ workers for async processing