Run Supabase 100% LOCALLY for Your AI Agents

---

---

AI Summary

Summary of Video Transcript

• Topic: Setting up Supabase locally for AI agents
• Supabase:
  • Open-source platform with PostgreSQL under the hood.
  • Offers features like authentication, object storage, and the PG Vector extension for vector databases.
  • Useful for managing conversation history, state, and vector databases for AI agents.

Detailed Instructions & Tips

• Local AI Package:
  • Includes Supabase for database and vector store capabilities.
  • Originally contained n8n, Ollama, Quadrant, and PostgreSQL.
  • Added Flow-wise, Open Web UI, and now Supabase, replacing PostgreSQL.
  • Quadrant remains for faster vector store use cases.
• Prerequisites:
  • Python, Git/GitHub Desktop, Docker/Docker Desktop.
  • Clone the dedicated GitHub repository for the local AI package.
• Environment Setup:
  • Create an .env file with variables for n8n and Supabase.
  • Set Postgres password, dashboard login, and Pooler tenant ID.
  • Generate JWT secret, anonymous key, and service role key using Supabase’s self-hosting guide.
• Docker Compose:
  • Use a separate Docker compose file for Supabase due to multiple containers.
  • The Python script combines Supabase’s Docker compose with the local AI package.
  • Containers run on the same Docker Network for easy connection.
• Starting the Local AI Stack:
  • Use provided commands based on architecture (Nvidia GPU, AMD GPU, CPU, or Mac).
  • The start-services.py script handles cloning, updating, and restarting containers.
  • Data in n8n workflows and Supabase tables persists across container restarts.
• Troubleshooting:
  • Common issues and solutions are provided in the README.

Creating an AI Agent

• n8n Workflow:
  • Automatically imports a local rag AI agent workflow.
  • Set up credentials for AMA and Supabase (Postgres node and vector store).
  • For vector store, use host.docker.internal as the host.
• Local File to Vector Store Workflow:
  • Watches for file changes in a local folder.
  • Cleans the vector database for updated files.
  • Extracts text from files and inserts vectors into Supabase.
• Testing the Agent:
  • Use the n8n workflow to add local files to the Supabase vector store.
  • Chat with the local AI agent to retrieve information from the vector store.

Conclusion

• The local AI package now includes Supabase, enhancing its capabilities for local AI development.
• The video provides a step-by-step guide to setting up and using the package.
• Future plans include expanding the package and incorporating community feedback.

URLs and Commands

• GitHub repository for the local AI package: (URL not provided in the transcript)
• Supabase self-hosting guide: (URL not provided in the transcript)
• Commands for starting the local AI stack, updating containers, and setting environment variables are mentioned but not provided in the transcript.