AI Chat Assistant for Kommo (AmoCRM) Integration - n8n Workflow

• CRM Administrators and Sales Operations teams using Kommo.


• Businesses needing 24/7 automated support in their chat channels.


• Developers looking for robust n8n templates integrating Langchain and OpenAI.


• Users seeking an efficient n8n workflow for voice message transcription and automated reply generation.

Managing high volumes of customer chats, including voice messages, can overwhelm support teams. This advanced n8n workflow solves this by instantly deploying an intelligent AI assistant directly within the Kommo chat environment. Not only does this n8n automation respond immediately to text queries using a sophisticated AI agent, but it also automatically handles voice messages by transcribing them via an OpenAI n8n node before generating a relevant response. This ensures rapid response times, frees up human agents for complex issues, and provides a customizable, context-aware experience, all powered by a robust n8n setup.

The entire process is initiated by the Webhook n8n trigger named 'newmessage' when a customer sends a chat message in Kommo.

1. Data Retrieval: The workflow immediately executes an HTTP Request n8n node ('getentity') to fetch contextual data related to the conversation.


2. Flow Control: A Switch n8n node ('hasStopTag') checks if the message contains a special tag that signals the AI to stop responding, allowing human agents to take over if necessary.


3. Media Check: The If n8n node ('isVoice') determines the message type. If it is a voice message, an HTTP Request ('get voice') downloads the audio file.


4. Transcription (If Voice): The audio file is passed to the 'transcribe voice' Langchain OpenAI n8n node, which uses Whisper to convert the voice recording into text.


5. AI Preparation: Regardless of the source, the message text is standardized using the 'setText' n8n node.


6. AI Response Generation: The core processing happens in the 'ai' Langchain Agent n8n node. This agent utilizes the 'model' (Chat OpenAI) and 'memory' (Buffer Window) n8n nodes to ensure context-aware, persistent conversation history. The agent uses this context to formulate a meaningful response.


7. Delivery: Finally, the n8n workflow uses two consecutive HTTP Request n8n nodes ('Get token' and 'Recieve message') to authenticate with the Kommo API and send the AI-generated reply back to the customer chat.

To use this powerful n8n workflow, follow these steps:

1. Import the n8n Template: Copy the provided JSON data and paste it directly into your n8n instance via the 'New' menu -> 'Import Workflow from JSON'.


2. Webhook Setup (n8n Trigger): Activate the 'newmessage' Webhook n8n node. Copy the resulting production URL and configure it as the outbound Webhook URL within your Kommo chat settings (or system responsible for passing chat data to n8n).


3. Credential Configuration:

1. Agent Customization: Review the parameters within the 'ai' n8n node to tailor the agent's personality, tools, and system prompts to match your desired support tone and knowledge base.

• newmessage (Webhook n8n Trigger): The starting point for the n8n workflow. Triggers upon receiving a new chat message payload from Kommo.


• getentity (HTTP Request n8n node): Used to retrieve supplementary details about the chat context or related CRM entity (e.g., deal ID, contact info) from Kommo.


• hasStopTag (Switch n8n node): A crucial flow control n8n node. It directs the workflow based on the presence of a user-defined tag, preventing AI response if human intervention is flagged.


• isVoice (If n8n node): Determines if the received message object contains a voice note attachment, branching the n8n workflow accordingly.


• transcribe voice (Langchain OpenAI n8n node): Downloads and converts voice notes to text using OpenAI's transcription model, preparing the input for the AI agent.


• ai (Langchain Agent n8n node): The core AI processor in this n8n template. It uses the input text, configured language model ('model'), and conversation history ('memory') to generate the final, intelligent reply.


• model (Langchain ChatOpenAI n8n node): Defines the specific LLM (e.g., GPT-4) used by the AI agent for generating high-quality responses.


• memory (Langchain Memory BufferWindow n8n node): Stores recent conversation history, allowing the AI agent in this n8n workflow to maintain context across multiple messages.


• Get token & Recieve message (HTTP Request n8n nodes): Handle API authentication and the final transmission of the AI-generated response back into the Kommo chat interface.