Azure AI extension in Azure Database for PostgreSQL - Flexible Server
APPLIES TO: Azure Database for PostgreSQL - Flexible Server
Azure Database for PostgreSQL flexible server extension for Azure AI enables you to use large language models (LLMS) and build rich generative AI applications within the database. The Azure AI extension enables the database to call into various Azure AI services including Azure OpenAI and Azure Cognitive Services simplifying the development process allowing seamless integration into those services.
Enable the azure_ai
extension
Before you can enable azure_ai
on your Azure Database for PostgreSQL flexible server instance, you need to add it to your allowlist as described in how to use PostgreSQL extensions and check if correctly added by running SHOW azure.extensions;
.
Tip
You might also want to enable the pgvector
extension as it is commonly used with azure_ai
.
Then you can install the extension, by connecting to your target database and running the CREATE EXTENSION command. You need to repeat the command separately for every database you want the extension to be available in.
CREATE EXTENSION azure_ai;
Note
To remove the extension from the currently connected database use DROP EXTENSION azure_ai;
.
Installing the extension azure_ai
creates the following three schemas:
azure_ai
: principal schema where the configuration table resides and functions to interact with it.azure_openai
: functions and composite types related to OpenAI.azure_cognitive
: functions and composite types related to Cognitive Services.
The extension also allows calling Azure OpenAI and Azure Cognitive Services.
Configure the azure_ai
extension
Configuring the extension requires you to provide the endpoints to connect to the Azure AI services and the API keys required for authentication. Service settings are stored using following functions:
Permissions
Your Azure AI access keys are similar to a root password for your account. Always be careful to protect your access keys. Use Azure Key Vault to manage and rotate your keys securely.
To manage service keys used by the extension, users require the azure_ai_settings_manager
role granted to them. The following functions require the role:
- azure_ai.set_setting
- azure_ai.get_setting
The azure_ai_settings_manager
role is by default granted to the azure_pg_admin
role.
azure_ai.set_setting
Used to set configuration options.
azure_ai.set_setting(key TEXT, value TEXT)
Arguments
key
Name of a configuration option. Valid values for the key
are:
azure_openai.endpoint
: Supported OpenAI endpoint (for example,https://example.openai.azure.com
).azure_openai.subscription_key
: A subscription key for an OpenAI resource.azure_cognitive.endpoint
: Supported Cognitive Services endpoint (for example,https://example.cognitiveservices.azure.com
).azure_cognitive.subscription_key
: A subscription key for a Cognitive Services resource.
value
TEXT
representing the desired value of the selected setting.
azure_ai.get_setting
Used to obtain current values of configuration options.
azure_ai.get_setting(key TEXT)
Arguments
Key
Name of a configuration option. Valid values for the key
are:
azure_openai.endpoint
: Supported OpenAI endpoint (for example,https://example.openai.azure.com
).azure_openai.subscription_key
: A subscription key for an OpenAI resource.azure_cognitive.endpoint
: Supported Cognitive Services endpoint (for example,https://example.cognitiveservices.azure.com
).azure_cognitive.subscription_key
: A subscription key for a Cognitive Services resource.
Return type
TEXT
representing the current value of the selected setting.
azure_ai.version
azure_ai.version()
Return type
TEXT
representing the current version of the Azure AI extension.
Examples
Set the Endpoint and an API Key for Azure OpenAI
select azure_ai.set_setting('azure_openai.endpoint','https://<endpoint>.openai.azure.com');
select azure_ai.set_setting('azure_openai.subscription_key', '<API Key>');
Get the Endpoint and API Key for Azure OpenAI
select azure_ai.get_setting('azure_openai.endpoint');
select azure_ai.get_setting('azure_openai.subscription_key');
Check the Azure AI extension version
select azure_ai.version();
Permissions
The azure_ai
extension defines a role called azure_ai_settings_manager
, which enables reading and writing of settings related to the extension. Only superusers and members of the azure_ai_settings_manager
role can invoke the azure_ai.get_settings
and azure_ai.set_settings
functions. In Azure Database for PostgreSQL flexible server, all admin users have the azure_ai_settings_manager
role assigned.
Upgrade the Azure AI extension
Newer versions of the extension can introduce new functionality and in-place upgrades of the extension are allowed. You can compare the currently installed version to the newest version allowed by using the SQL command:
SELECT * FROM pg_available_extensions
WHERE name = 'azure_ai'
To update an installed extension to the latest available version supported by Azure, use the following SQL command:
ALTER EXTENSION azure_ai UPDATE;
Share your suggestions and bugs with the Azure Database for PostgreSQL product team.
Related content
- Integrate Azure Database for PostgreSQL - Flexible Server with Azure Cognitive Services.
- Generate vector embeddings in Azure Database for PostgreSQL - Flexible Server with locally deployed LLM (Preview).
- Integrate Azure Database for PostgreSQL with Azure Machine Learning Services.
- Generate vector embeddings with Azure OpenAI in Azure Database for PostgreSQL - Flexible Server.
- Generative AI with Azure Database for PostgreSQL - Flexible Server.
- Recommendation System with Azure Database for PostgreSQL - Flexible Server and Azure OpenAI.
- Semantic Search with Azure Database for PostgreSQL - Flexible Server and Azure OpenAI.
- Enable and use pgvector in Azure Database for PostgreSQL - Flexible Server.