Take Odoo product management to a new dimension with artificial intelligence
Business management is evolving. Artificial intelligence is no longer limited to answering questions or generating content: it can now interact directly with business systems, query information, and execute operations using specialized tools.
In Bee Projects we are bringing this evolution to Odoo 19.0 through the development of an intelligent assistant connected to the ERP products, using artificial intelligence and an MCP (Model Context Protocol) server.
This integration allows users to interact with their product information using natural language, without the need to constantly navigate through different menus or know in advance where each piece of data is located within Odoo.
The goal is to transform the way companies query, understand, and manage their product catalog.
What is an intelligent assistant connected to Odoo?
It is a solution that connects an artificial intelligence model with the functionalities of Odoo through an MCP server.
Unlike a conventional chatbot, which only answers questions based on general information or a previously provided context, this assistant can use specific tools to query real information from the ERP.
For example, a user can make queries such as:
How many products do I have registered?
Search for products related to hot beverages.
Show me the details of this product.
What is the selling price of this item?
What physical and available stock do you have?
What suppliers are associated with this product?
What fields can I consult or modify in the products?
Artificial intelligence interprets the request, identifies what tools it needs, and uses the MCP server to access the corresponding functionalities.
How does the integration between AI, MCP, and Odoo 19 work?
The architecture is based on three main components:
1. Artificial intelligence
The AI model acts as a communication interface with the user.
It interprets questions, identifies intent, and decides what tools it needs to use to obtain an answer.
The solution includes different providers and artificial intelligence services, allowing work with alternatives such as:
Google Gemini.
Groq.
Models compatible with OpenAI.
GLM.
SiliconFlow.
Other compatible providers through specific configurations.
This allows adapting the integration to different models, capabilities, and needs of each project.
2. MCP Server (Model Context Protocol)
The MCP server acts as a bridge between artificial intelligence and Odoo.
Its responsibility is to expose tools that allow the model to interact with the available functionalities of the ERP.
Instead of allowing AI to access the database indiscriminately, specific tools, permissions, and access levels are defined.
This provides a more organized, controlled, and extensible architecture.
3. Product integration module
On this architecture, a specialized module is developed for interaction with Odoo products.
This component incorporates tools to query, search, analyze, and propose modifications to the catalog information.
In this way, product-related capabilities can evolve independently of the main MCP server.
Intelligent functionalities for Odoo products
The integration incorporates different tools that allow working with catalog information.
Product query and search
The assistant can query the catalog and retrieve information about products registered in Odoo.
Among the available functionalities are:
Query the total number of products.
Search for products by different criteria.
Obtain detailed information about a product.
Query prices and descriptions.
Retrieve inventory information.
Query associated suppliers.
Discover the available fields of the product model.
Semantic search using artificial intelligence
One of the most interesting functionalities is semantic search.
Traditional searches often rely on exact matches between words. However, users do not always know the exact name of a product or the terms used in its description.
Therefore, we are incorporating a semantic search system that allows finding products based on their meaning.
For example, if a user searches:
"Products related to hot beverages".
The system can find products that conceptually belong to that category, even if their names do not exactly contain those words.
How is this achieved?
Semantic information associated with the products is generated using artificial intelligence.
Vector representations (embeddings) are created using compatible models.
The results are stored in an independent semantic index.
When the user makes a query, its meaning is compared with the representations of the products.
Products with the highest semantic similarity are retrieved.
This process allows building a more flexible and contextual search engine, especially useful in catalogs where users use terms different from the commercial names of the items.
Additionally, the system is designed to progressively process the catalog, avoiding the need to generate all the semantic information at once.
Product modifications with control and confirmation
Artificial intelligence can not only query information: it can also help prepare modifications to the products.
For example, a user could request:
Change the selling price of a product.
Update its commercial description.
Modify descriptive information.
Prepare an update of certain fields.
To maintain proper control over these operations, modifications are managed through proposals pending confirmation.
The operation is as follows:
The user requests a modification using natural language.
The AI identifies the product and the fields involved.
A proposal is prepared with the requested changes.
The proposal is recorded as pending.
The operation requires confirmation before applying the changes.
This mechanism allows leveraging the capabilities of artificial intelligence without relinquishing control over business data.
The AI proposes; the user maintains control of the operation.
Security and control of tools
A business integration needs more than responsiveness: it also needs control mechanisms.
Therefore, the architecture includes a permission system to determine which tools the artificial intelligence can use.
Operations are organized at different levels:
Read: information query tools.
Write: operations that require modifying information.
Advanced: functionalities that require additional permissions.
Additionally, the tools can be configured independently, allowing functionalities to be enabled or restricted according to the needs of each installation.
It is important to highlight that the permissions of the MCP tools complement, but do not replace, the security and access control mechanisms inherent to Odoo.
A modular architecture ready to grow
One of the main objectives of development is to maintain a decoupled architecture.
The MCP server provides the communication infrastructure and tool management, while specialized modules incorporate the functionalities of each area of Odoo.
This separation allows:
To incorporate new tools without constantly modifying the main server.
To develop independent integrations for different Odoo models.
To keep product functionalities separate from other areas.
To incorporate new artificial intelligence providers.
To progressively expand the capabilities of the assistant.
To facilitate the maintenance and evolution of the ecosystem.
The same philosophy can be extended to other business areas, such as sales, purchases, inventory, accounting, CRM, and website.
What advantages does it bring to companies?
The combination of artificial intelligence and Odoo allows changing the way users interact with their ERP.
Among its main advantages are:
1. Interaction through natural language
Users can request information using common questions, without needing to know the technical structure of Odoo.
2. Contextual access to information
AI can use tools to retrieve real data from the system and generate responses based on them.
3. Smarter searches
Semantic search allows locating products by their meaning and context, not just by text matches.
4. Increased productivity
Navigation and manual query tasks to obtain catalog information are reduced.
5. Controlled operations
Modifications can be prepared as proposals that require confirmation.
6. Technological flexibility
Integration allows the use of different providers and artificial intelligence models.
7. Possibility of expansion
The modular architecture makes it easier to incorporate new capabilities and tools in other areas of Odoo.
The future of Odoo: business systems that understand what you need
The integration of artificial intelligence in ERPs represents a significant change in the way of working with business systems.
It is no longer just about incorporating a chatbot into an application, but about building assistants capable of interpreting requests, querying information, and using tools to interact with business processes.
With this solution for Odoo 19.0, at Bee Projects we are moving towards an ecosystem where artificial intelligence becomes a new interface for working with the data and functionalities of the ERP.
The MCP server provides the connection, specialized modules provide the tools, and artificial intelligence models allow interpreting user needs.
Odoo stops being just a system that users must consult to become a platform with which they can also interact through artificial intelligence.
Bee Projects: development and integration of artificial intelligence with Odoo
At Bee Projects, we work on the development of customized solutions for Odoo, business integrations, automations, and applications based on artificial intelligence.
Our approach combines modern technologies, modular architecture, and specialized tools to enhance the capabilities of business systems.
Does your company need to integrate artificial intelligence with Odoo?
Discover how we can help you develop intelligent solutions tailored to your business processes.
Bee Projects — Technology, automation, and artificial intelligence to take Odoo further...