This article will explain in detail how the standard template works. n8n for use with ByteGPT, as well as its functionalities and tools to assist in the integration of Kommo on IA.
Introduction
This automation manages the receipt of messages, the intelligent processing via AI and automatic data updates in CRM Kommousing n8n as orchestrator.
1. Trigger and Input
The flow uses a Chat Trigger as an initialization trigger. In this block, it is possible to find the Chat URLthe connection with the ByteGPT This is done using the link found in this step.
- Message Received: The flow starts when a new message event is detected in Kommo.
See in more detail how to connect the n8n to the Kommo at this link: How to connect n8n to Kommo using the ByteGPT.
Msgchat (Manual): An initial processing node that organizes the content of the received message so that the AI can interpret it correctly.
2. Main AI Agent
This is the "brain" of the flow. The node. AI Agent – Responses It uses a language model to decide how to respond to the user based on the context.
In this block it is possible to add the System Message to adjust the instructions and behaviors of the AI.
AI features used:
For the agent to function accurately, the block uses three important items:
AI Models (Chat Model): In the flow, the following is used: Gemini Chat ModelHe is responsible for logical reasoning and text generation.
Possible alternatives: GPT-4o, Claude 3.5 Sonnet or local models via Don't.
Memory (Simple Memory): Essential for maintaining the continuity of the conversation. It stores the latest interactions so that the AI knows what was said previously, preventing the user from having to repeat information.
- This feature uses a predefined window to store the latest interactions of IA, in this case, 5.
Tools (RAG): They allow AI to perform external actions or query databases. For example:
Get Google Sheets: A query tool that allows AI to read data from a spreadsheet (RAG – Retrieval-Augmented Generation) to respond with up-to-date information regarding prices, inventory, or services.
3. Modifying fields using the endpoint of ByteGPT
- HTTP Request (Update Fields) KommoInstead of constructing a complex JSON request body for the API native of Kommo, the is used ByteGPT as a facilitator.
In this block, a method is implemented. GET to an endpoint of UPDATE do ByteGPTThis endpoint receives simple parameters and the ByteGPT takes charge of translate this instruction For the CRM, updating lead or contact fields.
It is also possible to use specific endpoints for change of stage lead or break.
See more details in the article: How to update fields Kommo using ByteGPT + n8n
4. Completion
- Define Return (Manual): A final node that formats the response that will be sent back to the user's chat, ensuring that the communication loop is completed with the message generated by the AI.
Benefits of this structure:
Low Code (low code): the use of ByteGPT removes the technical barrier of authentication and structuring of JSON payloads from Kommo.
Dynamic Context: The combination of Memory and Google Sheets allows for personalized service based on real data.
Scalability: It's easy to replace the model (e.g., swapping Gemini for GPT) without needing to rebuild the CRM connection logic.
Bonuses
A Bytebio can assist in building flows for integration of IA using the ByteGPT with n8n in a punctual service.