Use Vectors with Autonomous AI Database

Autonomous AI Vector Database is optimized for vector workloads and designed for AI-powered applications. You can build semantic search, retrieval-augmented generation (RAG) pipelines, store and retrieve vector embeddings, and create agentic applications.

When to use Autonomous AI Vector Database

Use Autonomous AI Vector Database when you want to store vector embeddings and build AI-powered applications that retrieve content by meaning rather than by keyword alone.

For example, use it to:

In practice, a knowledge-search application can find related enterprise documents by meaning, a RAG application can ground LLM responses in proprietary content, and an AI agent can retrieve relevant context as it performs tasks. You can also build recommendation applications that match users, products, or content by similarity.

For more examples and use cases, see the Vector Database documentation.

How Vector Database works with Autonomous AI Database

Autonomous AI Vector Database is an Autonomous AI Database workload optimized for vector-based AI applications. It provides managed vector storage, indexing, embedding-model management, and vector similarity search for data such as documents, images, video, and audio.

Provision an Autonomous AI Vector Database by selecting Vector Database as the Autonomous AI Database workload type. The Vector Database workload requires Oracle AI Database 26ai or later.

Each Autonomous AI Database instance with vector workloads includes a fully managed vector database. From the Autonomous AI Vector Databases page, you can view instance information, monitor the database, manage tags, open Vector Database Console, and manage database users.

Before you Begin

Before using Autonomous AI Vector Database, verify the following:

Access Autonomous AI Vector Database

To access an Autonomous AI Vector Database instance:

  1. Open the Oracle Cloud Infrastructure Console.
  2. From the navigation menu, select Oracle Database, and then select Autonomous AI Database.
  3. On the Autonomous AI Databases page, locate the database that you created when provisioning the Autonomous AI Vector Database instance.
  4. Select the database display name.
  5. On the Autonomous AI Database details page, under Instances, select the Vector instance name.

The Autonomous AI Vector Databases page opens.

From the Autonomous AI Vector Databases page, you can:

For detailed steps, see Access an Autonomous AI Vector Database Instance.

Work with Autonomous AI Vector Database

Use Vectors with the Autonomous AI Transaction Processing (ATP) Workload

Use the Autonomous AI Transaction Processing (ATP) workload for applications that are primarily transactional or require more than 20 GB of non-vector data. To transition an Autonomous AI Vector Database instance to ATP, change the workload type to Autonomous AI Transaction Processing.

To identify user-created non-vector objects that count toward the 20 GB limit, see DBA_NONVECTOR_OBJECTS View. For the steps to change the workload type, see Upgrade an Autonomous AI Vector Database Instance for Production.

Tip:

Use the DBMS_VECDB_UTIL.GET_USER_NONVECDB_SIZE_GB function to obtain the total size of user-created non-vector database objects in gigabytes:

SELECT DBMS_VECDB_UTIL.get_user_nonvecdb_size_gb;

The function returns the total size as a NUMBER value in gigabytes. You can use the returned value to monitor non-vector data usage and determine when to change the workload type to Autonomous AI Transaction Processing.

Autonomous AI Vector Database limitations and restrictions