Odoo 20 makes the integration of AI into its core much tighter. The platform allows you to use assistants that will answer your questions, skills that define the assistant's behaviour, knowledge bases that you can search, and AI-based fields that can be created via Studio. It is natural for users from the Community edition to ask if this functionality is available for them. This article provides an answer to that question.
The short answer
Odoo 20 AI is not included in the Community edition. In summary:
Each ai_* module is included in the enterprise-20.0 package and is licensed with OEEL-1 (Odoo Enterprise Edition License) rather than the LGPL-3 license for Community.
To use these modules, an Enterprise subscription is necessary. Additionally, self-hosted versions of Odoo require the PostgreSQL server with the pgvector extension.
As there is no AI module in the standalone Community database, there is nothing to install. For the community itself, the pgvector extension is required only if you use Enterprise AI or any third-party vector module.
Where is the AI code located?
The source code of Odoo is split into two repositories. The first is called Community, and it has the server and the LGPL applications. The other repository is called Enterprise and has additional add-ons only. In order to launch Enterprise, one has to start the Community server and add the Enterprise folder to the --addons-path option.
In Odoo 20, the AI features are located in the Enterprise repository:
| Layer | Technical Name | Purpose |
| Core AI | ai | Agents, skills, sources, and connections to AI providers |
| Automatic Installer | ai_auto_install | Installs AI features when the environment supports them |
| Studio Integration | web_studio_ai_fields | AI-powered fields created from Studio |
| App-Specific AI | Other ai_* modules | AI features inside individual apps |
Since they are licensed under the OEEL-1 license, they are proprietary modules. They cannot be redistributed, and their code must not be used in a Community environment.
The core models: agents, skills and sources
Three models make up the base of the AI framework. Learning about them helps you understand what Community users will not be able to access.
ai.agent
An agent is a configured AI assistant. An agent has a name, instructions that set its role and tone, a chosen language model, and a set of tools that an agent can use. For example, you could create a sales assistant, an HR policy assistant, and a helpdesk assistant each with its assistant instructions.
ai.skill
A skill is a repeated ability that can be connected to an agent. By putting all the directions in one long message, you can create a task, like summarizing a ticket conversation or writing a follow-up quote, and use it with many agents. This keeps the directions clear and easier to check.
ai.agent.source
A source is a collection of knowledge connected to an agent, like files or attachments. The agent looks through these sources to find information before giving an answer. This method is called retrieval-augmented generation (RAG). The information is split into parts, turned into numbers called embeddings, and saved so the useful parts can be found fast.
Why is pgvector required?
RAG depends on similarity search over embeddings. Odoo keeps these embeddings inside your PostgreSQL database. Odoo does this by using the vector column type that comes from the pgvector extension. Because of this, you do not need a vector database.
Odoo's official documentation says that pgvector is needed to use the AI features. It also says that pgvector works with PostgreSQL 15 and newer. Community guides for Odoo 20 suggest using PostgreSQL 16 or newer and Python 3.12 or newer when you install a system. Because of these recommendations, you should plan your platform carefully.
On Ubuntu or Debian, you can install pgvector using this command:
# Replace 16 with your PostgreSQL version
sudo apt install postgresql-16-pgvector
After that, enable the extension in your Odoo database by running:
CREATE EXTENSION IF NOT EXISTS vector;
Important note on permissions
You must have superuser privileges to create the extension. If your Odoo application connects to PostgreSQL using a user, you might see an error saying "permission denied to create extension vector". This happens because regular users cannot create extensions. To fix it, make sure to create the vector extension one time as a superuser in your Odoo database. Do this before you try to install the AI modules. That way, everything will work as expected.
How does ai_auto_install work?
The ai_auto_install module installs the AI features automatically. Only when pgvector is present on the database server. If pgvector is missing, ai_auto_install will skip the installation step, and the rest of the database will keep working as usual.
In practice, this means:
- Enterprise with pgvector: ai_auto_install installs the AI features automatically.
- Enterprise, without pgvector: the database works as usual. Ai_auto_install does not install the AI features. If the AI menus are missing after an installation or upgrade, check pgvector first.
- Community: there is nothing to install automatically because the modules are not part of the addons path.
To check if pgvector is installed on your server, please run this query:
SELECT name, default_version, installed_version
FROM pg_available_extensions
WHERE name = 'vector';
If no row is returned, pgvector is not installed on the server. If a row is returned but installed_version is empty, pgvector is installed. The extension has not yet been enabled in your database.
AI fields in Studio
The web_studio_ai_fields module links the AI framework with Odoo Studio. It lets you create a field that gets its value from an AI prompt using details on the same record. Common examples are sorting a support ticket by taking a value from text or making a summary.
Two things are important to remember:
Studio is an enterprise application, so this feature needs two Enterprise parts: Studio itself and the AI framework.
AI fields use AI provider resources. It is useful to think about whether they should work when needed or every time a record is saved to control expenses and speed.
Hosting considerations
Whether you can use AI depends on how and where Odoo's hosted.
Odoo Online
Odoo manages the infrastructure, including the PostgreSQL layer. There is nothing for you to install. AI availability depends on your Enterprise plan. Odoo's own rollout rather than on your server configuration. This is the option.
Odoo.sh
Odoo.sh is Enterprise hosting with Git-based deployment and a managed PostgreSQL database. An Enterprise subscription is required. As you do not manage the database server directly, we recommend confirming pgvector support for your project and the PostgreSQL version with Odoo support or in the Odoo.sh documentation before planning an AI rollout.
On-premises or your own cloud
As you manage the environment yourself, please make sure the following requirements are met:
- Enterprise code in your --addons-path and a valid subscription.
- PostgreSQL 15 or later (16 or later is recommended) with pgvector installed.
- Superuser access to create the pgvector extension, as described above.
- Outbound internet access to your AI provider, if you use an external model.
- For Docker, a PostgreSQL image that includes pgvector.
For backups and upgrades, please note that a database with vector columns can only be restored on a server where pgvector is also installed.
Options for Community users
If you use the Community edition and want to add AI, you can choose from the following options.
- Upgrade to Enterprise
This is the choice. You receive all AI features, including agents, skills, sources, and Studio AI fields, and Odoo provides updates.
- Use third-party modules
Some open-source modules add AI chat, knowledge bases, and vector storage to Odoo. Before you use any of them, check that the modules support Odoo 20 and that they are actively maintained.
- Build a small custom integration
If you need a custom module that can call an AI service from server actions. Common examples are summarizing a conversation, drafting a reply, or classifying leads. Keep API keys in system parameters by putting them in the code.
- Connect an external automation tool
Automation tools can work alongside Odoo. Automation tools seem to read records through API calls, send them to an AI service, and write the results back. Automation tools work with any edition. Leave Odoo unchanged. Automation tools add another system to maintain.
A point to avoid
Do not copy OEEL-1 modules into a Community installation. This action is not allowed by the license. The modules rely on code that is only available in Enterprise.
If you need AI features that are supported, choose Enterprise. If Enterprise does not include AI, check pgvector. Then, check the PostgreSQL version. After that, verify whether ai_auto_install ran. If you host Odoo on your server, install the pgvector plan in the superuser step and test your backups. If you use Community, a custom integration is suitable; for a larger knowledge base, third-party modules may be suitable, but check them carefully.
Odoo 20 AI is an Enterprise feature set. Odoo 20 AI is licensed under OEEL-1. The core parts of Odoo 20 AI are ai.agent, ai.skill, and ai.agent.source. The parts ai.agent, ai.skill, and ai.agent.source are added to Studio. Rely on pgvector to retrieve knowledge. Community users do not receive the parts ai.agent, ai.skill, and ai.agent.source of Odoo 20 AI. There are still choices. Third-party modules and custom integrations can meet needs.
To read more about Odoo 20 Roadmap Explained: Key Updates and Future Expectations, refer to our blog, Odoo 20 Roadmap Explained: Key Updates and Future Expectations.