n8n Autonomous Customizable Support Chatbot for Intercom/Discord Thread Reports

This n8n workflow automates the process of handling incoming support requests, leveraging AI to understand the intent, retrieve relevant information, and generate appropriate responses or actions. It's designed to create a more efficient and intelligent support system.

Description

This workflow acts as an autonomous, customizable support chatbot. It receives incoming messages (likely from a webhook, suggesting integration with platforms like Intercom or Discord), processes them using an AI agent, and then routes the output to either generate a response or perform a specific action, such as creating a report. The AI agent utilizes a chat model, memory, and a Pinecone vector store for contextual understanding and information retrieval.

What it does

1. Receives Incoming Data: The workflow is triggered by an incoming webhook, expecting a payload containing a support request or message.
2. Initial Data Preparation: It sets specific fields on the incoming data, likely to standardize the input for the AI agent.
3. AI Agent Processing: The core of the workflow, an AI Agent, processes the input. This agent is configured with:
   • OpenAI Chat Model: To understand natural language and generate human-like responses.
   • Simple Memory: To maintain conversational context within a session.
   • Pinecone Vector Store: To retrieve relevant information from a knowledge base, enabling the chatbot to answer questions based on pre-existing data.
   • OpenAI Embeddings: Used by the Pinecone Vector Store for semantic search.
4. Output Filtering: The AI Agent's output is then filtered based on a condition, likely to differentiate between a direct AI response and a command/action.
5. Conditional Routing (Switch): Depending on the filtered output, the workflow routes the data to different branches:
   • True Branch (AI Response): If the AI determines a direct response is needed, it proceeds to generate and send that response.
   • False Branch (Action/Report): If the AI detects a need for an action (e.g., creating a report), it proceeds down this path.
6. Action/Report Generation:
   • HTTP Request: Sends an HTTP request, likely to an external system (e.g., a reporting tool, another API) to create a report or perform a specific action.
   • Intercom Integration: Integrates with Intercom, suggesting it can create or update conversations, or send messages within Intercom based on the AI's determination.
7. Data Aggregation and Splitting: The workflow includes Aggregate and Split Out nodes, which are likely used to manage data structures, potentially combining or separating information before or after processing by the AI or external services.
8. Final Response/Action: The workflow concludes by either sending an AI-generated response back (likely via the initial webhook or a platform-specific node like Intercom) or completing an external action.

Prerequisites/Requirements

• n8n Instance: A running n8n instance to host the workflow.
• Webhook Endpoint: An external system (e.g., Intercom, Discord, a custom application) configured to send messages/requests to the n8n webhook URL.
• OpenAI API Key: For the OpenAI Chat Model and Embeddings.
• Pinecone Account and API Key: For the Pinecone Vector Store (requires a pre-populated Pinecone index with your knowledge base).
• Intercom Account and Credentials: If using the Intercom node for communication.
• External API/Service Credentials: Any credentials required for the HTTP Request node if it interacts with another external service for reporting or other actions.

Setup/Usage

1. Import the Workflow: Import the provided JSON into your n8n instance.
2. Configure Webhook:
   • Activate the "Webhook" node and copy its URL.
   • Configure your external system (e.g., Intercom, Discord bot, custom application) to send POST requests to this URL with the relevant message payload.
3. Set up Credentials:
   • OpenAI: Create an OpenAI credential in n8n and link it to the "OpenAI Chat Model" and "Embeddings OpenAI" nodes.
   • Pinecone: Create a Pinecone credential in n8n and link it to the "Pinecone Vector Store" node. Ensure your Pinecone index is set up and populated with the data you want the AI to reference.
   • Intercom: Create an Intercom credential in n8n and link it to the "Intercom" node.
4. Customize AI Agent:
   • Review the configuration of the "AI Agent" node, including the prompt, tools, and other settings, to align with your specific support needs and knowledge base.
   • Adjust the "OpenAI Chat Model" settings (e.g., model, temperature) as needed.
5. Configure Logic Nodes:
   • Review and adjust the conditions in the "Filter" and "Switch" nodes to accurately route messages based on the AI Agent's output and your desired workflow.
6. Customize HTTP Request (if applicable):
   • If the "HTTP Request" node is used for reporting or other actions, configure its URL, method, headers, and body to match the API of your target service.
7. Activate the Workflow: Once all configurations are complete, activate the workflow.

The workflow will now listen for incoming messages, process them through the AI agent, and respond or take action accordingly.