3 Get Started with Oracle AI Data Platform

Oracle AI Data Platform helps you quickly build, orchestrate, and manage end-to-end data workflows across your organization. This guide walks you through the essential steps to create your first AI Data Platform and start working with data in a governed and scalable environment.

Topics:

Requirements for AI Features

In order to use AI features in Oracle AI Data Platform, you need to provision an Oracle Autonomous AI Lakehouse instance.

As part of creating your AI Data Platform instance, an Oracle Autonomous AI Lakehouse must be created and configured. You can choose to create the Autonomous AI Lakehouse instance as part of creating your first AI Data Platform instance or choose to use an existing Autonomous AI Lakehouse instance. If you are using an existing Autonomous AI Lakehouse, you must also provide the password for the ADMIN user of that instance.

Note:

Existing Autonomous AI Lakehouse instances must be at least 26ai or above. When you create a new Autonomous AI Lakehouse instance as part of creating your AI Data Platform instance, it is automatically configured as 26ai or above.

The Autonomous AI Lakehouse instance serves as the foundational data layer for AI workloads within AI Data Platform and is used as the default vector store for all AI-native capabilities. The Autonomous AI Lakehouse instance requires Oracle Autonomous AI Lakehouse version 26ai or later and is mandatory for enabling AI features in AI Data Platform, including:

  • Knowledge Bases (KB),
  • Agent Flows,
  • and AI Compute.
Without an Autonomous AI Lakehouse instance, AI Data Platform continues to support structured data ingestion and transformation, but AI-native capabilities remain disabled.

Note:

If your tenancy has reached one or more Autonomous AI Lakehouse quotas, you can't create a new Autonomous AI Lakehouse instance when creating your AI Data Platform instance. In that case, you must either select an existing Autonomous AI Lakehouse instance or contact your support team to request a quota increase.

You must have the following IAM policies enabled for AI features:

Allow Oracle AI Data Platform service to call LLMs hosted on the OCI GenAI service:

allow any-user to use generative-ai-family in tenancy where all { request.principal.type='aidataplatform'}

Note:

Depending on when your AI Data Platform was created, your principal type may be 'datalake' instead of 'aidataplatform'. You can check your principal type by viewing the details of your AI Data Platform OCID.
Allow Oracle AI Data Platform service to provision and manage Oracle Autonomous AI Lakehouse resources to enable AI features:
allow group <aidpAdminGroup> to manage autonomous-databases in <aidp compartment>

Storage Encryption Options

You can use Oracle-managed encryption keys, or you can choose your own customer-managed key (CMK) from OCI Vault when your organization requires direct control over key lifecycle and rotation.

Oracle AI Data Platform encrypts data at rest by default. When you configure your first AI Data Platform instance, you can choose to keep the default Oracle-managed encryption keys, or you can choose your own CMK from OCI Vault when your organization requires direct control over key lifecycle and rotation. Use Oracle-managed keys when you want the simplest setup and don't need to manage key material yourself. Use customer-managed keys when your security policy requires you to control the vault, key rotation process, access boundaries, or audit trail for the encryption key.

Customer-Managed Key Prerequisites (Preview)

Before you select a customer-managed key, make sure the following prerequisites are in place:

  • You have permission to view OCI Vaults and keys in the target compartment. For more information configuring vaults and keys in OCI, see Configuring IAM Policies.
  • You have an OCI Vault available in the same region as your AI Data Platform deployment.
  • You have a master encryption key in that vault.
  • The key is enabled and available for use.
  • You must choose an AES symmetric master key with a 256-bit length.

Default Behavior – Oracle Managed Keys

If you don't select a customer-managed key, AI Data Platform uses Oracle-managed keys automatically. This is the default option and requires no additional vault configuration. Choose this option when:

  • You want the fastest provisioning path.
  • Your team doesn't need to manage or rotate the encryption key directly.
  • Your security policy allows Oracle-managed encryption for service data at rest.

Considerations Before Choosing Customer-Managed Keys

Customer-managed keys provide more control, but they also add operational responsibility.

  • If the vault or key becomes unavailable, encryption-dependent operations can fail.
  • If the key is disabled, deleted, or scheduled for deletion, access to protected data can be impacted.
  • Key rotation, compartment design, and IAM policies become part of your operational runbook.
  • Provisioning or update flows can take longer than the default Oracle-managed path.

Troubleshooting

If you don't see the customer-managed key option when setting up your AI Data Platform instance, check the following:

  • Confirm that the feature is enabled in your tenancy.
  • Confirm that you have permission to read vaults and keys.
  • Confirm that the vault and key are in the same region as the AI Data Platform instance.

Limitations

You cannot change your encryption keys after creating an AI Data Platform instance. In order to change encryption strategy, you must create a new AI Data Platform instance and move the code to that.

Create Your First Oracle AI Data Platform

Before you create your first Oracle AI Data Platform, you need to ensure you have the correct IAM permissions.

Tutorial icon LiveLabs Sprint

Before creating or using an AI Data Platform, check the following:

  • You have the appropriate permissions for the compartment where you want to create the AI Data Platform. For more information, see Managing Compartments.
  • If you already have tag-defaults set for your compartment you may face issues while creating an AI Data Platform instance in that scope. For more information, see Known Issues.
  1. In the Home section of OCI, click Analytics & AI.
  2. Click AI Data Platform.

    AI Data Platform screen with the Create AI Data Platform button highlighted

  3. Click Create AI Data Platform. AI Data Platform performs a check to validate that all required IAM policies are in place.
  4. Provide a name and description for your AI Data Platform instance.
  5. Provide a name and description for your first AI Data Platform workspace.
  6. Select whether your AI Data Platform instance uses a public or private network configuration.
    1. For private networks, you choose a compartment and VCN where Private Service Access (PSA) is already enabled. You then select a subnet and one or more network security groups.
    2. If your data sources are in a different private network, you need to select that option and provide the VCN, subnet, and network security groups for that private network as well.
    3. If your data sources require access through a SCAN proxy, require advanced DNS configuration, or your private network enforces a network source, expand Advanced – optional, select the appropriate options and provide the required settings.
  7. An Oracle Autonomous AI Lakehouse instance is required for AI features. Select whether to create a new instance or use an existing instance.
    • For Create new, provide a new password for the ADMIN user of the instance.
    • For Choose existing, select the compartment and instance and enter the password for the ADMIN user of the existing instance. Your existing instance must be 26ai or above.
  8. To associate an Oracle Analytics Cloud instance with your AI Data Platform, select whether to use an existing Analytics Cloud instance or create a new one at the same time as your AI Data Platform instance.
    1. For Choose existing, select the compartment, provide a domain name, and select the existing instance. Provide the vault secret OCID as the password.
    2. For Create new, provide the domain name and vault secret OCID.
    3. Select None to associate an Analytics Cloud instance later.
  9. Select whether encryption keys are Oracle-managed (default setting) or customer-managed.
    • Oracle-managed key: This is the default option and requires no additional vault configuration.
    • Customer-managed key: Select if your organization requires direct control over key lifecycle and rotation. You need to provide the Vault OCID and Key OCID for your encryption key. Click Validate to confirm the provided OCIDs are valid.

    Storage encryption step of Create AI Data Platform. Customer-managed key is selected.

  10. Select one of the following from Add Policies:
    • Standard (Recommended): Applies access settings broadly at the tenancy level.
    • Advanced: Allows you to configure fine-grained access at the compartment level.
  11. Review the missing policies, if any, and apply them as necessary.
  12. Expand Optional Policies. Review and add any additional policies your instance requires. For more information, see IAM Policies for Oracle AI Data Platform.
  13. Click Advanced Options to provide metadata tags for your AI Data Platform. Click Add tag to add multiple.
  14. Click Create AI Data Platform.

Access an Oracle AI Data Platform

You can access any Oracle AI Data Platform that you or a role you belong to have been provided permissions for.

You require at least USE permissions from the AI Data Platform IAM policies to access an AI Data Platform. For more information, see IAM Policies for Oracle AI Data Platform.
  1. In the Home section of OCI, click Analytics & AI.
  2. Click AI Data Platform.
  3. Click the AI Data Platform you want to access.

Access Oracle AI Data Platform from a URL

You can access your Oracle AI Data Platform instance directly from a URL.

The URL for your Oracle AI Data Platform uses the following format, where tenant_name is the name of the OCI account you use to log in and domain_name is the domain you select: 

https://<hash>.datalake.oci.oraclecloud.com/#?&tenant=<<tenant_name>>&domain=<<domain_name>>

You can also get a direct link to your instance by right-clicking your AI Data Platform instance name on the home page and clicking Copy Link.

You can share your AI Data Platform instance URL with other users. Those users need to have appropriate permissions to access the instance. For more information, see About Permissions.

Understanding the Workspace and UI Flow

Once you've created or accessed your Oracle AI Data Platform, the user interface helps you navigate and manage all key components seamlessly inside the instance.

Left Navigation Overview

When inside an AI Data Platform, the left sidebar contains:

  • Home - Brings you back to the dashboard
  • Master Catalog - View all registered data assets across workspaces
  • Workspace - Direct access to workspace details
  • Select Workspace - Drop-down to switch between different workspaces
    • Inside a workspace, you’ll see:
      • Workflow - Design and schedule data workflows
      • Compute - Manage Spark compute environments
  • Data Sharing - Share data with other services supporting Delta Share protocol
  • Auto populate catalog - Automatically ingest metadata from connected sources
  • Notifications - Search and filter system messages you receive
  • Roles - Configure access control (RBAC)
  • Audit logs - Search and filter the history of objects in your AI Data Platform

Next Steps After Setup

Once you have created your Oracle AI Data Platform, you can prepare it for regular use.

When you have your AI Data Platform created and have accessed it for the first time, these are the next steps to ensure its ready for use:

  • Review IAM Policies and Roles - Ensure correct access is set up for users and groups via the Roles tab. For more information, see Roles.
  • Set Up Compute - Configure Spark compute in the Compute tab to run your workflows efficiently. For more information, see Compute.
  • Ingest and Organize Data - Start by creating a Catalog to organize your data sources. For more information, see Data Management.
  • Explore Notebooks - Use notebooks for ad hoc exploration, transformations, or machine learning tasks. For more information, see Notebooks.

Edit an Oracle AI Data Platform

You can edit the details of an Oracle AI Data Platform instance you own.

  1. In the Home section of OCI, click Analytics & AI.
  2. Click AI Data Platform.
  3. Click Actions three dot iconActions next to your AI Data Platform and click Edit.
  4. Modify the details and click Save.

Delete an Oracle AI Data Platform

You can delete unused or redundant Oracle AI Data Platform instances you own.

  1. In the Home section of OCI, click Analytics & AI.
  2. Click AI Data Platform.
  3. Click Actions three dot icon Actions next to your AI Data Platform and click Delete.
  4. Select whether to delete:
    • All managed data in the Master Catalog
    • All files and folders in all workspaces
  5. Enter Delete where prompted. Click Delete.

Session Management

Oracle AI Data Platform enforces session management controls to ensure a secure and reliable user experience.

The session lifetime for AI Data Platform is 8 hours. User sessions automatically expire after this session lifetime. This behavior aligns with the Oracle Cloud Infrastructure (OCI) security standards and reduces the risk of unauthorized access by ensuring that session tokens are not valid indefinitely.

After 8 hours, users must re-authenticate to continue working in the instance.

These controls are designed to balance usability with strong security and governance practices across the platform.