DocuChat: Instant Document-to-Chatbot Conversion

Inspiration

We noticed creators wasting hours formatting content for chatbots when AI could automate this. Existing solutions required technical skills or expensive subscriptions. Our "lightbulb moment" came when a teacher friend said: "I just want my PDF syllabus to answer student questions automatically."

What It Does

  • Converts PDFs/docs into shareable chatbots in <30 seconds
  • Generates public links that work on any device
  • Uses document context to answer questions accurately
  • Built entirely on Supabase for real-time data handling

How We Built It

Tech Stack:

  • Frontend: React.js + Tailwind CSS
  • Backend: Supabase (Auth, Storage, PostgreSQL)
  • AI: OpenAI API with Supabase Edge Functions
  • Hosting: Netlify + Supabase Storage

Key Implementation:

  1. Supabase Auth for instant account creation
  2. File uploads to Supabase Storage with RLS policies
  3. Document processing via Supabase Edge Functions

Challenges We Ran Into

  1. PDF Extraction: Solved with Supabase Storage triggers
  2. Rate Limiting: Implemented row-level security in PostgreSQL
  3. Cold Starts: Optimized Supabase Function memory allocation

Accomplishments

  • Achieved quick response using Supabase caching
  • Scaled to multiple concurrent users during testing
  • Maintained high uptime during hackathon presentation
  • Implemented full CI/CD pipeline in short time

What We Learned

  • Supabase's real-time capabilities enabled instant updates
  • PostgreSQL row-level security simplified permissions
  • Edge Functions were perfect for document preprocessing
  • The importance of proper bucket policies in Storage

What's Next

  1. Supabase-powered team collaboration features
  2. Fine-tuned open-source LLMs via Supabase
  3. Usage analytics with Supabase Postgres reporting
  4. Mobile app using Supabase auth synchronization

Built With

Share this project:

Updates