AI Knowledge-Base Agent (RAG)

Inside this workflow

Overview

**What it automates:** Webhook (question + context) → OpenAI reads context, answers question, cites source + confidence → responds to caller with fallback "not in docs" if uncertain. This is the answer core; you add your own retrieval (vector search, BM25) in the step before. **Business outcome:** Support teams spend 20% of time on FAQ-answerable questions. This workflow is the grounded-answer engine that powers your knowledge-base chatbot or Slack integration. **Apps & nodes:** Webhook, retrieval provider (your DB/vector store), OpenAI (grounded answering), HTTP response **Setup (4 steps):** 1. Create webhook for questions 2. Build/integrate your retrieval step (vector search, Supabase, Algolia) 3. Add OpenAI API key with grounding instructions 4. Test with known questions **Difficulty:** Intermediate-Advanced (20 min setup) | **Credentials:** OpenAI, your retrieval provider

What's included

Quickstart

1. 01

   Download the JSON

   With an active Library Pass, download the workflow file right from this page or your dashboard.

2. 02

   Import into n8n

   Choose Import from File and select the downloaded JSON. The full agent graph appears, ready to configure.

3. 03

   Plug in credentials

   Each integration node prompts for credentials on first run. The setup guide lists every credential the agent expects.

4. 04

   Activate and test

   Run once with sample input, confirm the expected output, then flip the activate toggle.

Sign in to leave a review.