English Website Chatbot Webhook with OpenAI GPT-5 & Qdrant

Deploy an English Website Chatbot Webhook powered by OpenAI GPT-5 + Qdrant

This n8n workflow exposes a secured, header-authenticated webhook that turns your website content into an English chatbot. It uses OpenAI for intent handling and response generation, then retrieves relevant knowledge from Qdrant to return structured JSON search results—ready to plug into your site or chat client.

What this workflow does

• Secured webhook intake: Receives a POST request on a header-authenticated webhook, normalizes the payload into a chat message and session ID.
• Intent classification: Uses OpenAI to classify the message intent into smalltalk, search (knowledge query), or reject.
• Safety-first handling: For rejected or out-of-scope requests, returns a fixed safety message as JSON.
• Smalltalk replies: For smalltalk, uses OpenAI to generate a short English response and formats it into a simple JSON text response.
• Knowledge retrieval with Qdrant: For knowledge queries, searches a Qdrant vector collection (optionally reranked by Cohere) and compiles top unique sources (titles, URLs, and snippets) into context.
• Strict JSON result generation: An OpenAI-powered agent produces up to five relevant results plus an optional follow-up question, following a strict JSON schema; the workflow then parses and cleans the output.
• Webhook response: Returns the final JSON payload to the original caller.

Use cases

• Add an English website chatbot to route visitor questions to your knowledge base.
• Enable search-style answers with citations (title + URL + snippet) using Qdrant retrieval.
• Build an automation-friendly API endpoint for SaaS help centers and documentation sites.

Technical details

• n8n nodes: Webhook, Code, Switch, Sticky Note, n8nn8n-nodes-langchainagent, n8nn8n-nodes-langchainopen ai.
• Integrations: OpenAI GPT-5 (intent detection, smalltalk, answer generation) and Qdrant (vector search; optional Cohere reranking).
• Deploy by configuring header authentication, adding OpenAI credentials, and pointing to your populated Qdrant collection (e.g., website_knowledge).