AI Chat Agent with OpenAI, Memory, and Web Search

Overview

This workflow creates an intelligent AI chat agent that combines OpenAI's language model with conversational memory and real-time web search capabilities via SerpAPI. It is designed to power interactive chat applications where the AI can remember context across messages and fetch up-to-date information from the internet when needed. This is particularly useful for building customer support bots, research assistants, or any conversational interface that requires both knowledge retention and live data access.

The workflow is triggered by incoming chat messages and processes them through an AI agent that coordinates between the language model, memory buffer, and search tool. The agent can decide when to use web search to answer questions that require current information, while maintaining conversation history for coherent multi-turn dialogues.

Workflow Steps

1. Manual Trigger (No auth) — This node allows you to manually start the workflow for testing purposes. It has no parameters configured and simply provides a button to execute the workflow on demand.

2. When chat message received (No auth) — This is a webhook-based trigger that listens for incoming chat messages. It generates a unique webhook URL that you can integrate with your chat frontend or messaging platform. The options parameter is empty, meaning default webhook behavior is used.

3. AI Agent (No auth) — This is the core orchestrator node that manages the conversation flow. It coordinates between the OpenAI model, memory buffer, and SerpAPI tool. The agent decides which tool to use based on the user's query and maintains the conversation state.

4. OpenAI Chat Model (API Key auth) — This node connects to OpenAI's GPT-4o-mini model to generate responses. It requires an OpenAI API key configured in n8n credentials. The model parameter is set to gpt-4o-mini, which is a cost-effective and fast model suitable for chat applications. The options parameter is empty, using default model settings.

5. Window Buffer Memory (No auth) — This node implements a sliding window memory that stores recent conversation history. It allows the AI to remember context from previous messages within a configurable window size. The parameters are empty, using default memory settings.

6. SerpAPI (API Key auth) — This tool node enables web search functionality via SerpAPI (Google Search API). It requires a SerpAPI API key configured in n8n credentials. The options parameter is empty, using default search settings. The agent calls this node when it determines that real-time web data is needed to answer a query.

Setup Instructions

To use this workflow, you need the following accounts and credentials:

1. OpenAI Account: Sign up at platform.openai.com and create an API key. In n8n, create an "OpenAI" credential and paste your API key.

2. SerpAPI Account: Register at serpapi.com and obtain an API key. In n8n, create a "SerpAPI" credential with your key.

3. Webhook URL: After activating the workflow, copy the webhook URL from the "When chat message received" node. Integrate this URL into your chat application (e.g., a custom web chat widget, Slack app, or Telegram bot).

4. Optional: Configure the memory window size in the "Window Buffer Memory" node if you want to limit or extend conversation context.

Use Cases and Variations

This workflow can be adapted for multiple scenarios:

• Customer Support Bot: Replace SerpAPI with a knowledge base tool (e.g., vector store or database lookup) to answer product-specific questions.
• Research Assistant: Keep SerpAPI for web search and add additional tools like Wikipedia or news APIs.
• Multi-Platform Chat: Connect the webhook to different messaging platforms (Slack, Discord, Telegram) by adding appropriate trigger nodes.
• Personal Assistant: Add calendar or email tools to let the AI manage schedules and send messages.
• Language Translation: Replace the OpenAI model with a translation-specific model or add a translation tool node.

To extend the workflow, you can add more tool nodes (e.g., database queries, file operations) and connect them to the AI Agent. The agent will automatically learn to use new tools based on their descriptions.

{
  "meta": {
    "instanceId": "workflow-2528b000",
    "versionId": "1.0.0",
    "createdAt": "2025-09-29T07:07:42.167307",
    "updatedAt": "2025-09-29T07:07:42.167320",
    "owner": "n8n-user",
    "license": "MIT",
    "category": "automation",
    "status": "active",
    "priority": "high",
    "environment": "production"
  },
  "nodes": [
    {
      "id": "trigger-9ee3ef68",
      "name": "Manual Trigger",
      "type": "n8n-nodes-base.manualTrigger",
      "typeVersion": 1,
      "position": [
        100,
        100
      ],
      "parameters": {}
    },
    {
      "id": "939bb301-5e12-4d5b-9a56-61a61cca5f0d",
      "name": "OpenAI Chat Model",
      "type": "n8n-nodes-base.noOp",
      "position": [
        640,
        460
      ],
      "parameters": {
        "model": "gpt-4o-mini",
        "options": {}
      },
      "credentials": {
        "openAiApi": {
          "id": "8gccIjcuf3gvaoEr",
          "name": "OpenAi account"
        }
      },
      "typeVersion": 1,
      "notes": "This lmChatOpenAi node performs automated tasks as part of the workflow."
    },
    {
      "id": "372777e8-ce90-4dea-befc-ac1b2eb4729f",
      "name": "Window Buffer Memory",
      "type": "n8n-nodes-base.noOp",
// ... truncated (copy to see full JSON)