n8n Slack RAG Workflow: Supabase + Gemini + OpenRouter

Ask your product backlog PDF in Slack—automatically

With this n8n Slack RAG Workflow: Supabase + Gemini + OpenRouter, you can upload a product backlog PDF, index it into a Supabase vector store, and then query it instantly from Slack using a slash command. The workflow runs Retrieval-Augmented Generation (RAG): relevant passages are retrieved from Supabase using Google Gemini embeddings, and answers are drafted by an OpenRouter chat model.

What this workflow does

• Index a backlog PDF: Receives a form submission with a PDF upload, extracts text from the file, generates embeddings with Google Gemini (PaLM), and inserts the content into a Supabase vector store table.
• Initialize a knowledge base: Loads default documents and embeddings to set up the same Supabase knowledge base.
• Answer questions from Slack: Handles a Slack slash command request to the workflow webhook path /backlog-query and immediately posts an in-channel “search in progress” acknowledgment.
• Run RAG with retrieval + generation: Sends the user’s question to an OpenRouter-powered RAG agent, retrieves relevant backlog passages from Supabase using Gemini embeddings, and generates a Slack-formatted answer.
• Respond in the channel: Posts the final response back to the originating channel using Slack’s response_url.

Use cases

• Product teams want quick answers to “What’s the status of feature X?” directly in Slack, based on the backlog PDF.
• SaaS operators can centralize backlog knowledge for faster alignment across engineering and stakeholders.
• Automation engineers can deploy a ready-made RAG pattern using n8n, Supabase, Gemini, OpenRouter, and Slack.

Technical details

• n8n nodes: webhook, sticky note, form trigger, HTTP request, n8nn8n-nodes-langchainagent, and “extract from file”.
• Integrations: Supabase vector store (table: backlog_documents), Google Gemini (PaLM) for embeddings, OpenRouter for the chat model, and Slack slash commands.
• Setup highlights: Configure the form trigger (share it for PDF uploads), add Supabase + Gemini + OpenRouter credentials, create the Slack slash command POSTing to /backlog-query, and use the generated webhook URL in Slack.