A few days ago, I started to familiarize myself with API connections to different artificial intelligence models, exploring how far they could go and what uses they could have, especially when integrating them into customized Odoo environments.
Based on this experience, I set out to develop my own MCP server for Odoo 19, with the goal of exploring and expanding the integration possibilities that these technologies offer.
Objectives and results
The main objective is to develop a modular tool that, supported by artificial intelligence, allows connecting Odoo with different models and facilitates everyday work, both through specific instructions and through process automation.
Everything revolves around the MCP server
The MCP server is the central component of this architecture. It is responsible for managing communication with the different artificial intelligence models through their respective APIs, as well as centralizing the management of permissions and the prompts associated with the various modules that integrate with it.
Its function is to define what tools and operations are available for each interaction, establishing the limits of what AI can do within Odoo.
Through communication with the AI model, the server exposes tools that allow querying information, exploring the available fields in the Odoo models, and, when permissions allow, making modifications to the data.
In this way, AI is not limited to answering questions: it can interact with the ERP and execute specific operations within the limits defined by the permission system.
Minimizing costs
I have prioritized the integration of artificial intelligence models that offer free plans, suitable for moderate use and for testing various functionalities without necessarily incurring additional costs.
Currently, the server allows working with Gemini, Groq Cloud, and GLM, offering different alternatives based on the needs of each project.
However, if more advanced models or specific features available through paid plans are required, it is also possible to incorporate them into the architecture, adapting the configuration to the needs of each Odoo environment.
The most interesting part: custom integrations
One of the most interesting parts of the project is the possibility of developing additional tools that allow maximizing the capabilities of the MCP server and adapting them to specific needs.
These are some of the functionalities I have started to develop:
A project with many possibilities
One of the great advantages of this architecture is that it is not limited to current functionalities. It is possible to develop and incorporate new tools as needs arise, progressively expanding Odoo's capabilities and the possibilities of interaction with artificial intelligence models.
The real potential lies in adapting artificial intelligence to the actual needs of each company, expanding Odoo's capabilities beyond its standard functionality.
The potential is enormous and, precisely, that flexibility is one of the reasons why I believe it is worth continuing to explore this path.