Dynamic AI Agent for HighLevel CRM and Advanced Knowledge Retrieval - n8n Workflow

Agencies managing client communication and sales pipelines within HighLevel (GoHighLevel).
Developers looking for advanced n8n templates integrating LangChain agents with microservices.
Users needing a centralized AI assistant capable of performing CRM operations (tasks, contacts, opportunities) and retrieving complex, structured data.
Businesses requiring an intelligent communication funnel across Gmail, Slack, and Telegram, powered by a sophisticated n8n node setup.

This is one of the most sophisticated n8n templates for enterprise automation, designed to centralize customer relationship management and knowledge retrieval. The core of this solution is the GHL orchestrator, an n8n LangChain Agent that acts as a central AI assistant. This agent listens for input from multiple sources (Gmail, Slack, Telegram) using dedicated n8n trigger nodes and decides which action to take: perform a CRM operation (like updating a contact in HighLevel) or retrieve complex contextual knowledge.

The value provided by this n8n workflow lies in its modularity (using n8n's Microservice Communication Protocol - MCP) and its contextual depth. The agent can access real-time web data via the Perplexity n8n node, query Google Sheets using a specialized n8n node, and retrieve vectorized personal data stored in Supabase, providing highly accurate and context-rich responses back to the user via the original input channel. Every critical function is exposed through a specialized n8n node or sub-workflow, ensuring scalability and maintainability.

This complex n8n workflow begins with several possible entry points, each serving as an n8n trigger.

1. Trigger & Input: Incoming messages via the Gmail Trigger, Slack Trigger, Telegram Trigger, or a LangChain When chat message received n8n trigger are captured and funneled through the Edit Fields n8n node.


2. Orchestration: The data flows to the GHL orchestrator (an n8n Agent node) powered by the GPT 5 (ChatOpenAI) LLM. This agent maintains conversation context using a Chat Memory n8n node (Postgres).


3. Tool Selection (CRM): If the user asks to manage CRM data (e.g., "Create a task in HighLevel"), the orchestrator utilizes the MCP Client GHL n8n node, which routes the request to the dedicated HighLevel microservice (GHL MCP server). This n8n workflow server exposes multiple HighLevel n8n node functions (like creating opportunities or updating contacts).


4. Tool Selection (Knowledge): If the user asks a question requiring external information (e.g., "What are the Q3 sales figures?"), the orchestrator uses the MCP Client Knowledge n8n node, which consults the Knowledge MCP server. This n8n template combines multiple retrieval systems: real-time web search, Google Sheets data, and RAG retrieval against Supabase vectors (using the Embeddings OpenAI and Reranker Cohere n8n nodes for high-quality context).


5. Tool Selection (Document Context): If the user needs context from a specific document, the orchestrator uses the document retrieval n8n node, which searches Google Drive, downloads the file, and then uses file processing n8n nodes (e.g., Extract from PDF1, Analyse Image1) to gain file context.


6. Final Response: Once the GHL orchestrator determines the optimal response, the result is sent back through the appropriate communication channel n8n node (Gmail, Slack, or Telegram). This ensures a full-circle automated response within the single n8n workflow.

To set up this comprehensive n8n workflow, follow these steps:

1. Import the n8n template: Copy the provided JSON and import it into your n8n instance via the Workflows section (Import from JSON).


2. Configure Credentials: Establish credentials for all integrated services. You will need credentials for HighLevel (required for all CRM n8n node operations), OpenAI (required for the LLM and vision/audio n8n nodes), Postgres (for Chat Memory), Google Drive/Sheets, and Supabase (for the vector store n8n node).


3. Set up Triggers: Configure the webhook URLs for the Slack Trigger, Telegram Trigger, and When chat message received n8n trigger nodes on their respective platforms. The Gmail Trigger will require a secure connection and configuration.


4. Activate MCP Servers: Ensure all three MCP Server nodes (GHL MCP server, Knowledge MCP, Google Drive MCP Server) are active. The orchestrator n8n node relies on these sub-workflows being operational.


5. Activate and Test: Activate the main n8n workflow. Test by sending a message to one of the configured channels (e.g., Slack) requesting a HighLevel task update or a complex knowledge query. The entire system uses the n8n node structure to manage interaction.

GHL orchestrator (LangChain Agent n8n node): The core decision-making unit. It uses tools to execute HighLevel CRM actions and knowledge retrieval. Key Configuration: Uses GPT 5 (ChatOpenAI) as the Language Model and Chat Memory (Postgres) for stateful conversations.
Gmail Trigger / Slack Trigger / Telegram Trigger (n8n trigger nodes): Serve as the entry points for the automation, capturing incoming communications. Function: Initiates the n8n workflow upon receiving a new message or email.
HighLevelTool n8n node (e.g., Create or update a contact in HighLevel): Central to the CRM automation, exposed via the GHL MCP server. Function: Executes specific HighLevel API calls (contacts, tasks, opportunities, appointments) as directed by the AI agent.
Vectorized personal data (Supabase VectorStore n8n node): Used within the Knowledge MCP to perform Retrieval-Augmented Generation (RAG). Function: Queries the Supabase vector database for contextual information, enhanced by Embeddings OpenAI and Reranker Cohere n8n nodes for precise retrieval.
Real time web search (PerplexityTool n8n node): Provides the AI agent with current, external data. Function: Used by the Knowledge MCP to answer questions requiring live internet access.
Extract from PDF1 / Extract from CSV1 (ExtractFromFile n8n node): Used in the Google Drive microservice to process documents. Function: Extracts textual content from specific file types after they are downloaded via the Download File n8n node, allowing the AI to gain file context.
MCP Trigger / MCP Client Tool (n8n node family): Essential for modularity, allowing the agent to call functions exposed in separate sub-workflows. Function*: Enables the main orchestrator n8n node to interact reliably with the GHL, Knowledge, and Google Drive microservices.