No-Code RAG with n8n and Supabase
Supabase's vector extension
Supabase (a Postgres-based backend platform) offers a pgvector extension that turns a regular Postgres table into a vector store — letting you build RAG without a separate dedicated vector database service.
Building the pipeline visually in n8n
Document upload trigger → chunking node (Text Splitter) → OpenAI Embeddings node → Supabase insert node. For queries: user question → embed → Supabase vector search node → AI node with retrieved context → response.
When no-code RAG makes sense
This approach is ideal for internal tools, small knowledge bases, and non-developers who need a working RAG system fast — for high-scale or highly customized RAG, a code-based implementation (Lesson 5) offers more control.
Key Takeaways
- Supabase's pgvector extension turns Postgres into a vector store.
- n8n can build the full RAG pipeline visually — chunking, embedding, retrieval, generation.
- This is ideal for internal tools and non-developers needing RAG fast.
- Code-based RAG still offers more control for high-scale or custom needs.
Build a no-code RAG scenario
In n8n, build a scenario that chunks and embeds a document into Supabase, then a second scenario that answers a question using retrieved context.