AI-Powered API Data Fetcher

This n8n workflow demonstrates how to empower an AI agent to dynamically call an API to fetch data based on a user's chat message. It sets up a conversational AI that can understand requests for external data and execute an n8n sub-workflow (acting as a tool) to retrieve it, making your AI more dynamic and capable.

What it does

This workflow orchestrates an AI agent to respond to chat messages by potentially calling an external API via a sub-workflow:

1. Listens for Chat Messages: The workflow is triggered whenever a chat message is received.
2. Initializes AI Agent: It sets up an AI Agent with a conversational memory and an OpenAI Chat Model.
3. Defines an API Calling Tool: A "Call n8n Workflow Tool" is configured, allowing the AI to execute a specific sub-workflow (which in turn can make HTTP requests to external APIs) when needed.
4. Processes Chat Input: The AI Agent receives the incoming chat message and determines if it needs to use the defined tool to fetch data from an API.
5. Executes API Call (if needed): If the AI decides an API call is necessary, it triggers the "Call n8n Workflow Tool," which then executes the designated sub-workflow to make the HTTP request.
6. Formats API Response: The data fetched by the HTTP Request node is then processed by an "Edit Fields (Set)" node to structure the output.
7. Returns Result to AI: The result from the API call (or the AI's direct response if no API call was needed) is returned, enabling the AI to provide an informed answer.

Prerequisites/Requirements

• n8n Instance: A running n8n instance.
• OpenAI API Key: Configured as an n8n credential for the "OpenAI Chat Model" node.
• Sub-Workflow for API Call: A separate n8n workflow that performs the actual HTTP request to your desired API. This workflow needs to be callable by the "Call n8n Workflow Tool" and should be designed to accept parameters (if any) from the AI and return relevant data.
• Basic understanding of LangChain concepts: Familiarity with LLM chains, agents, tools, and memory will be beneficial for customization.

Setup/Usage

1. Import the Workflow: Import the provided JSON into your n8n instance.
2. Configure OpenAI Credentials:
   • Locate the "OpenAI Chat Model" node.
   • Select or create an OpenAI API credential.
3. Configure the "Call n8n Workflow Tool":
   • Locate the "Call n8n Workflow Tool" node.
   • Specify the ID of the sub-workflow that will handle your API calls. This sub-workflow should start with an "Execute Workflow Trigger" node.
   • Define the "Description" for the tool, clearly explaining its purpose to the AI (e.g., "Use this tool to fetch current weather data for a given city.").
   • Define the "Input Schema" for the tool, describing the expected parameters the AI should provide (e.g., { "city": { "type": "string", "description": "The city name" } }).
4. Create the Sub-Workflow (if not already existing):
   • Create a new n8n workflow.
   • Add an "Execute Workflow Trigger" node as its starting point.
   • Add an "HTTP Request" node to make the actual API call. Configure its URL, headers, and body based on the API you want to query. You can use expressions to dynamically insert parameters passed from the main AI workflow (e.g., {{ $json.city }}).
   • Add an "Edit Fields (Set)" node to format the API response into a clean output.
   • Ensure this sub-workflow returns the data in a format the AI can understand.
5. Activate the Workflow: Enable the workflow in n8n.
6. Test: Send a chat message (via the "Chat Trigger" node) to your AI that would require an API call (e.g., "What is the weather in London?"). Observe the execution to confirm the AI correctly identifies the need for the tool and executes the sub-workflow.