AI-Powered Email Response System with RAG and Pinecone - n8n Workflow

• Technical support teams looking to automate level 1 responses.


• Businesses needing to leverage large documentation sets (RAG) to generate context-aware email replies.


• Developers seeking robust n8n templates for advanced AI agent orchestration.


• Users who need a complete, scalable n8n solution for managing high-volume email queries.

Managing high volumes of incoming support or sales emails can be time-consuming, especially when answers rely on proprietary documentation. This comprehensive n8n workflow solves this by establishing a powerful AI Email Auto-Responder System.

This system features dual components: a document ingestion pipeline and an active response agent. The ingestion pipeline uses the Google Drive Trigger to monitor for new files, automatically splitting, embedding (using OpenAI), and indexing the content into a Pinecone vector database.

The primary email flow uses the Gmail Trigger to activate the AI agent. The agent first determines the intent of the email. If a complex query is detected, it utilizes Retrieval-Augmented Generation (RAG) tools connected to Pinecone to fetch the most relevant data before drafting a polished, context-aware reply. This highly functional n8n template dramatically reduces manual effort while maintaining high quality in correspondence.

This powerful n8n workflow operates in two main phases:

Phase 1: Knowledge Base Ingestion (RAG Setup)

1. Document Trigger: The Google Drive Trigger waits for new or updated documents to appear in a specified folder.


2. File Download: The Google Drive n8n node downloads the new file content.


3. Preparation: The content is processed by the Default Data Loader and segmented using the Recursive Character Text Splitter.


4. Embedding and Indexing: The Embeddings OpenAI1 n8n node generates vector embeddings, which are then stored in the Pinecone Vector Store node, creating the RAG knowledge base used by the AI agent.

Phase 2: AI Email Response Automation

1. Workflow Start: The Gmail Trigger is the entry point, firing whenever a new email arrives in the monitored inbox.


2. Intent Check: The flow uses the Set Data node to prepare the content, followed by the Check if Question n8n node (an LLM Chain utilizing an OpenAI Chat Model) to classify the incoming email's intent (Is it a question requiring a search?).


3. Conditional Routing: The If n8n node directs the flow. If the email is classified as a question (True branch), it proceeds to search and retrieval.


4. Information Retrieval (RAG Agent): The Search Information n8n agent node takes over. This sophisticated agent uses various tools:

1. Response Generation: Once the Search Information agent collects the required context, the flow passes to the Response writer n8n agent node. This agent uses another OpenAI Chat Model to synthesize the retrieved information and the original email into a professional, human-like response.


2. Final Action: The Gmail n8n node drafts and sends the final, AI-generated email reply to the original sender. This complete n8n workflow ensures quick and accurate handling of customer queries.

To deploy this powerful n8n workflow, follow these steps:

1. Import: Copy the entire JSON code and import it into your n8n instance via the 'New' menu -> 'Import from JSON'.


2. Credentials Setup: This n8n template requires several credentials. Set up connections for:

1. RAG Configuration: Configure the Google Drive Trigger to monitor the folder containing your documentation. Ensure the Pinecone index names match the configuration in the Pinecone Vector Store and Pinecone Vector Store1 n8n nodes.


2. Agent Customization: Review the system prompts within the OpenAI Chat Model n8n nodes to tailor the tone and personality of the AI agent writer.


3. Activate: Save the n8n workflow and set the execution mode to 'Active' to start monitoring your inbox.

Gmail Trigger (n8n trigger): Starts the main n8n workflow upon receiving a new email. Essential for defining the scope of automation.
Set Data (n8n node): Used early in the flow to standardize or extract necessary data fields (like sender, subject, and body) from the incoming Gmail Trigger output before AI processing.
Check if Question (Chain LLM n8n node): Utilizes an LLM to perform an intent classification on the email content, determining if it warrants a RAG lookup. Uses OpenAI Chat Model2.
If (n8n node): A core logic component that branches the flow based on the result of the intent check, optimizing resources by only running the RAG agent when necessary.
Search Information (LangChain Agent n8n node): The primary brain of the response system. This powerful n8n node orchestrates the use of multiple tools and memory to gather context and answers. It includes OpenAI Chat Model (LLM), Simple Memory, and RAG tools.
Answer questions with a vector store (LangChain Tool n8n node): The crucial RAG tool, configured with Pinecone Vector Store1 and Embeddings OpenAI, which performs the actual knowledge base lookup.
Response writer (LangChain Agent n8n node): Takes the context retrieved by the search agent and uses OpenAI Chat Model3 to draft the final, comprehensive email response.
Gmail (n8n node): Executes the final action, sending the AI-generated response back to the original sender.
Google Drive Trigger (n8n trigger): Starts the separate RAG ingestion n8n workflow when documents are updated or added, keeping the knowledge base current.
Pinecone Vector Store (n8n node): Used in the ingestion flow to save document chunks and their vectors, powered by Embeddings OpenAI1, ensuring the RAG data is always fresh.