AI-Powered MIS Agent with Dynamic Data Processing and RAG - n8n Workflow

Business Intelligence teams needing instant, conversational access to business data.
Automation developers seeking advanced examples of building AI agents within n8n templates.
Users requiring automated processing of structured data derived from incoming communication (using a Gmail trigger).
Organizations utilizing Pinecone and PostgreSQL for data storage and retrieval in a RAG architecture.

This is a comprehensive, modular n8n workflow designed to centralize and automate business intelligence tasks. It establishes an AI Agent capable of interacting with various business systems (Google Sheets, PostgreSQL, Google Drive) and responding to user queries received through a conversational interface (Chat trigger) or by classifying incoming communication via a Gmail trigger. By leveraging advanced LangChain nodes and proprietary custom code, this powerful n8n automation ensures data freshness via scheduled indexing into a Pinecone vector store, providing a robust solution for dynamic, data-driven decision making. Implementing this n8n template greatly enhances operational efficiency.

This complex n8n workflow is structured around three parallel entry points:

1. Email Ingestion and Classification (Gmail Trigger):
The workflow starts with the Gmail Trigger whenever a new email arrives. The content is immediately passed to the OpenAI Chat Model1 for processing by the Text Classifier n8n node. This classifier determines the email's intent (e.g., 'Daily Sales', 'Address Update', 'Customer Info') and routes the message via subsequent Set and If nodes to dedicated data handling branches, preparing data for downstream actions like saving attachments to Google Drive.

2. Conversational MIS Agent (Chat Trigger):
When a user sends a query (via the When chat message received n8n trigger), the workflow initializes the core MIS Agent1. This powerful agent, built on LangChain, uses Simple Memory1 for context and chooses from various tools: Google Sheets8 (for data lookup), Postgres (for database interaction), Calculator, or an internal Call n8n Workflow Tool1 (to execute other segments of this n8n workflow).

3. Data Operation Execution:
Regardless of the trigger source, requests requiring complex data manipulation pass through the Operation Classifier (powered by OpenAI Chat Model4). This determines if the request involves retrieving report folder URLs (Drive Folders URL1) or initiating a data retrieval/report generation process. If a report is needed, nodes interact with Google Sheets and use Code nodes to format the data, generating a final file (using Convert to File nodes) before uploading it back to Google Drive via the relevant Google Drive n8n node.

4. Knowledge Base Maintenance (Schedule Trigger):
A separate branch begins with a Schedule Trigger for routine maintenance. It identifies documents in the Knowledge_base Google Drive folder, uses the Default Data Loader and Recursive Character Text Splitter to chunk the data, generates embeddings using Embeddings OpenAI, and updates the Pinecone Vector Store with the latest vectorized data. This ensures the AI agent always has up-to-date context.

1. Download and Import: Save this n8n workflow JSON file and import it directly into your n8n instance.


2. Credential Setup: You must provide valid credentials for Gmail (for the initial n8n trigger), Google Drive, Google Sheets, PostgreSQL, OpenAI (for language models and embeddings), and Pinecone (for the vector store).


3. Configure AI Tools: Ensure all AI components, especially the various OpenAI Chat Model instances, are connected to the correct model endpoints and API keys.


4. Configure Internal Call: The Call n8n Workflow Tool1 must be configured correctly to reference the ID of this specific n8n workflow to enable the agent's ability to recursively trigger internal logic.


5. Set Drive Paths: Update the Set nodes (like Drive Folders URL and Drive Folders URL1) with the actual folder IDs from your Google Drive where reports and knowledge base documents are stored. Deploy the n8n trigger nodes after setup is complete.

Gmail Trigger: The primary n8n trigger for monitoring and initiating email processing based on new message arrival.
MIS Agent1 (LangChain n8n node): The core intelligence node that uses tools, memory, and an OpenAI Chat Model to interpret complex queries and execute actions.
OpenAI Chat Model (Multiple): Used extensively for classification (email intent, operation type) and as the main language model for the AI Agent.
Pinecone Vector Store (LangChain n8n node): Essential for maintaining the Retrieval Augmented Generation (RAG) knowledge base, allowing the agent to utilize context from indexed documents.
Google Sheets Tool / Postgres Tool: Specialized tool nodes enabling the MIS Agent to query external structured data sources directly, fulfilling data requests.
If / Switch Nodes: Crucial for complex routing logic, directing the flow based on classification output or data validation within this n8n workflow.

• Convert to File: An n8n node used in the reporting pipeline to transform processed data streams (like JSON outputs) into downloadable file formats (e.g., CSV or PDF) for storage in Google Drive.