Oracle AI Database Assistant

12 Oracle AI Database Assistant

Oracle AI Database Assistant enables you to interact with Oracle Enterprise Manager using natural-language questions to assess enterprise health, investigate operational issues, and gain insights from Enterprise Manager telemetry.

Note:

Starting with Enterprise Manager 24ai Release 1 Update 10 (24.1.0.10), Oracle AI Database Assistant is available in Enterprise Manager. This enhanced experience, which replaces the earlier Ask EM functionality, is described in this chapter. In this release, Oracle AI Database Assistant focuses primarily on telemetry-oriented interactions and Enterprise Manager operational data.

Oracle AI Database Assistant serves as a conversational interface for Enterprise Manager. You can ask natural-language questions and receive responses based on Enterprise Manager telemetry and monitoring information. Oracle AI Database Assistant sends only the user prompt and the context required to answer the request to the configured Large Language Model (LLM) provider.

Oracle AI Database Assistant can help you:

Retrieve target status and health information
Investigate incidents and operational issues
Review job execution activity
Analyze monitoring metrics
View interactive widgets and visualizations
Access AI-generated operational insights

Topics:

Upgrading from Ask EM to Oracle AI Database Assistant

Oracle AI Database Assistant replaces the Ask EM functionality that was available in earlier Enterprise Manager 24ai release updates.

If you are upgrading from an earlier Oracle Enterprise Manager 24ai release update that uses Ask EM, review the following changes before using Oracle AI Database Assistant:

Oracle AI Database Assistant no longer depends on OCI Ops Insights for chatbot functionality. You do not need an OCI Ops Insights subscription, OCI Ops Insights-enabled host or database targets, or OCI Ops Insights onboarding solely to use Oracle AI Database Assistant in Enterprise Manager.
Oracle AI Database Assistant supports multiple LLM provider options, including the Enterprise Manager provider and bring-your-own LLM providers. For bring-your-own LLM configurations, you can use the OpenAI provider option for OpenAI-compatible endpoints or the Ollama provider option for Ollama endpoints. Oracle AI Database Assistant uses LLM credentials to manage provider connections.
Oracle AI Database Assistant integrates directly with OCI Generative AI when you use the Enterprise Manager provider.
Oracle AI Database Assistant provides enhanced interactive widget-based responses.

To get started with Oracle AI Database Assistant, upgrade to Enterprise Manager 24ai Release 1 Update 10 (24.1.0.10) or later. For information on configuring and using Oracle AI Database Assistant after the upgrade, see the following sections in this chapter.

Prerequisites for Oracle AI Database Assistant

Before you configure Oracle AI Database Assistant, make sure the required connectivity, credentials, and access requirements are in place.

General Prerequisites

Before you configure Oracle AI Database Assistant, make sure that you:

Log in to Enterprise Manager as a Super Administrator.

Note:
It is recommended that the same Super Administrator create the required LLM credential, own the associated credentials, and complete the Oracle AI Database Assistant configuration, including user access control tasks.
Review the supported LLM providers and the corresponding provider-specific prerequisites:
- Enterprise Manager (recommended): Uses Oracle Cloud Infrastructure (OCI) Generative AI through Enterprise Manager.
- OpenAI: Uses a customer-managed LLM provider endpoint configured through the OpenAI provider option in Oracle AI Database Assistant. This option can be used for OpenAI-compatible endpoints.
- Ollama: Uses a customer-managed Ollama endpoint configured through the Ollama provider option in Oracle AI Database Assistant.
Verify that Oracle AI Database Assistant is enabled in Enterprise Manager. Oracle AI Database Assistant is enabled by default and to disable it, set the Oracle Management Service (OMS) property enable_genai_assistant to false using the following command:
```
emctl set property -name enable_genai_assistant -value false
```
To re-enable Oracle AI Database Assistant, run the following command:
```
emctl set property -name enable_genai_assistant -value true
```

Additional Prerequisites for the Enterprise Manager LLM Provider

If you use the Enterprise Manager provider, make sure you have:

An Oracle Cloud account in one of the following supported OCI regions:
- Japan Central (Osaka): ap-osaka-1
- Germany Central (Frankfurt): eu-frankfurt-1
- Brazil East (Sao Paulo): sa-saopaulo-1
- UK South (London): uk-london-1
- US Midwest (Chicago): us-chicago-1
Note:
Oracle AI Database Assistant is currently supported only in the regions listed above. Support for additional OCI Generative AI regions will be added in future updates.

The following OCI policies so that Enterprise Manager can access the OCI Generative AI service:

allow group <your-group-name> to manage opsi-genai-em-session in compartment <your-compartment-name>
allow group <your-group-name> to use generative-ai-family in compartment <your-compartment-name>

An OCI Named Credential for communication with OCI. You can create the named credential in advance in Enterprise Manager or while creating the LLM credential. For information on creating a named credential outside the LLM credential flow, see Define a Global Named Credential in Enterprise Manager for OCI.
To create an OCI Named Credential, you need:
- The following OCI tenancy and API signing key information:
  - Tenancy OCID
  - User OCID
  - Public Key Fingerprint
  - Private Key
  - Private Key Passphrase
  For information on how to obtain these details, see Required Keys and OCIDs.
- Access to an OCI Generative AI region. For a list of supported regions, see Generative AI Regions.

If you use the Enterprise Manager provider, Oracle AI Database Assistant uses the configured LLM credential to communicate with the LLM service hosted in OCI. Requests originate in the browser and are authenticated by the Enterprise Manager OCI authentication signing service before being sent to OCI. As a result, an OCI Named Credential is required when you use the Enterprise Manager provider.

When you use the Enterprise Manager provider, Oracle AI Database Assistant includes a complimentary allocation of AI tokens for OCI Generative AI models through the OCI Ops Insights service. After the complimentary allocation is exhausted, usage charges are billed to your OCI account based on the selected Generative AI model and the OCI region in which the service is hosted. For more information, see OCI Generative AI documentation and OCI Generative AI pricing.

Additional Prerequisites for Bring-Your-Own LLM Providers

If you use OpenAI or Ollama, make sure you have:

An API key, if the LLM provider requires one to authenticate the connection
The name of the model that you want to use
The URL for the provider endpoint
Network connectivity between Enterprise Manager and the provider endpoint

If you use the OpenAI or Ollama provider option, Oracle AI Database Assistant uses the configured LLM credential to communicate with the selected customer-managed LLM provider. The OpenAI provider option can be used for OpenAI-compatible endpoints. The provider can reside on premises or in the cloud. When using these providers, OMS communicates directly with the configured provider endpoint. Make sure OMS can establish outbound network connectivity to the endpoint and that any required firewall ports are open. If outbound connections must traverse a corporate proxy, configure the required proxy settings.

Roles, Privileges, and Access to Oracle AI Database Assistant

Grant the required role and privileges to allow access to Oracle AI Database Assistant.

To access Oracle AI Database Assistant, users must have the required role and privilege assignments. The required access depends on whether you are a Super Administrator or a non-super administrator, and on the credential configuration used by Oracle AI Database Assistant.

Note:

The following subsections describe the access requirements for Super Administrators and non-super administrators to access and use Oracle AI Database Assistant. To simplify user access setup, it is recommended that the Super Administrator who configured Oracle AI Database Assistant click the Set up Oracle AI Database Assistant icon (

) in the upper-right corner of the chat panel, open the Oracle AI Database Assistant Setup wizard, and add users on the User Access Control page. This setup flow simplifies access management by allowing the configuring Super Administrator to grant selected users Oracle AI Database Assistant access, including the credential access required by the assistant, from a single page.

Super Administrator

A Super Administrator can configure Oracle AI Database Assistant and access the Oracle AI Database Assistant icon in the UI. However, a Super Administrator may still be unable to use Oracle AI Database Assistant if another Super Administrator created the configured LLM credential.

The following privileges are required for a Super Administrator to configure and use Oracle AI Database Assistant. Note that a Super Administrator may already have some of these privileges by default:

View any OMS configuration property privilege on the OMS Configuration Property resource type (VIEW_ANY_OMS_PROPERTY)
View Credential on the configured LLM credential
View Credential on the configured OCI Named Credential when using the Enterprise Manager provider

Example Scenario

A Super Administrator sees the Oracle AI Database Assistant icon enabled in the UI. However, if the configured LLM credential was created by another Super Administrator, Oracle AI Database Assistant displays an access error in the chat panel. In this case, the Super Administrator must contact the credential owner to request access, or create an LLM credential that they can access.

Non-super Administrator

A non-super administrator must be granted the EM_ASKEM_ADMIN role by a Super Administrator to access Oracle AI Database Assistant. If this role is not assigned, the Oracle AI Database Assistant icon is disabled.

In addition to the EM_ASKEM_ADMIN role, the following privileges are required for a non-super administrator to use Oracle AI Database Assistant:

View any OMS configuration property privilege on the OMS Configuration Property resource type (VIEW_ANY_OMS_PROPERTY)
View Credential on the configured LLM credential
View Credential on the configured OCI Named Credential when using the Enterprise Manager provider

Example Scenario

A non-super administrator who is not granted the EM_ASKEM_ADMIN role cannot access Oracle AI Database Assistant, and the assistant icon is disabled. If the role is granted but the required privileges are not granted, Oracle AI Database Assistant may open, but an access error is displayed in the chat panel. In this case, the non-super administrator must contact the Super Administrator who configured Oracle AI Database Assistant to request the required access.

Configuring Oracle AI Database Assistant

Configure Oracle AI Database Assistant in Enterprise Manager by creating an LLM credential for the provider that you want to use.

An LLM credential defines how Oracle AI Database Assistant connects to the LLM provider that you want to use. During the LLM credential creation process, you provide the connection and authentication details needed to send requests and receive responses from the model.

To create an LLM credential:

Log in to Enterprise Manager as a Super Administrator.
From the navigation menu, select Setup, then Security, and then LLM Credentials.
In the LLM Credential Setup page, click + Add.
In the Create LLM Credential wizard, review the information in the Overview step and click Next.
In the LLM Credential step, provide the following details for interaction with the LLM:
1. Name: Enter a name for the credential.
2. Provider: Select the LLM provider that you want to use:
  - Enterprise Manager: If you select Enterprise Manager, provide the following information:
    1. OCI Named Credential: Select an existing OCI Named Credential to communicate with OCI.
    2. Region: Select one of the supported OCI regions.
    3. Compartment OCID: This value is automatically populated based on the selected OCI Named Credential.
    If you do not already have an OCI Named Credential, click + Create OCI Named Credential and provide the required OCI authentication details.
  - OpenAI: If you select OpenAI, which can be used for OpenAI-compatible endpoints, provide the following information:
    1. API Key: This is only required if the LLM provider requires one to authenticate the connection.
    2. Model: Enter the name of the model that Oracle AI Database Assistant should use to process requests.
    3. URL: Enter the endpoint URL that Enterprise Manager uses to connect to the selected LLM provider.
  - Ollama: If you select Ollama, provide the following information:
    1. API Key: This is only required if the LLM provider requires one to authenticate the connection.
    2. Model: Enter the name of the model that Oracle AI Database Assistant should use to process requests.
    3. URL: Enter the endpoint URL that Enterprise Manager uses to connect to the selected LLM provider.
Click Test Connection to verify the LLM provider connection.

For the Enterprise Manager provider, Test Connection validates connectivity from both OMS and the browser. For OMS to connect to the Enterprise Manager LLM, the required proxy configuration must be in place. For the browser to connect, the browser must be able to access OCI. For the OpenAI and Ollama provider options, Test Connection verifies that OMS can communicate with the configured provider endpoint.
Click Save to save the LLM credential.

The LLM Credential Setup page is displayed, and the newly created LLM credential is listed in the LLM Credentials list.

Managing LLM Credentials

From the LLM Credentials list, you can perform the following actions for an existing credential:

Update: Update the LLM credential to modify the authentication details used to connect to the LLM provider.
Set as Preferred LLM Credential: Set the credential as the preferred LLM credential used by Oracle AI Database Assistant for LLM communication. The first LLM credential that you create is automatically assigned as the preferred LLM credential. Oracle AI Database Assistant uses the preferred LLM credential as its system-wide credential. After a preferred LLM credential is set, it is displayed on the LLM Credential Setup page.
Delete: Delete the selected LLM credential.

Using Oracle AI Database Assistant

Use Oracle AI Database Assistant to ask natural-language questions and view widget-based responses in Enterprise Manager.

Oracle AI Database Assistant provides a single-pane chat experience in which you can submit natural-language questions and receive conversational responses and interactive widget-based results. Oracle AI Database Assistant is in preview mode, and Oracle is constantly working to improve the accuracy and quality of its GenAI-generated responses.

To access Oracle AI Database Assistant:

In Enterprise Manager, click the Oracle AI Database Assistant icon () in the upper-right corner of the console.
Oracle AI Database Assistant opens as an overlay on the right side of the current page.

Note:

If the Oracle AI Database Assistant icon is disabled or if an access error is displayed in the chat panel, review the earlier sections in this chapter for information about prerequisites, access requirements, and configuration.

Oracle AI Database Assistant Interface

Oracle AI Database Assistant uses a single-pane interface that provides introductory guidance and suggested question patterns to help you get started with supported queries related to metrics, incidents, jobs, and other Enterprise Manager monitoring data. Enter your question in the chat panel to request information and receive a relevant widget or visualization. For example, a question such as Show a list of job executions in the last 24 hours returns a jobs table widget.

Chat Controls

Oracle AI Database Assistant provides the following controls in the chat panel:

Set up Oracle AI Database Assistant (): Open the Oracle AI Database Assistant Setup wizard. In the setup wizard, you can review the configured LLM credential settings and, when applicable, grant selected users access to Oracle AI Database Assistant on the User Access Control page.
Maximize and Minimize Oracle AI Database Assistant ( and ): Expand Oracle AI Database Assistant to view more content or reduce it to fit your current workspace.
End Oracle AI Database Assistant Session (): Close Oracle AI Database Assistant and end the current chat session.

Interactive Widget-based Responses

When you ask a question, Oracle AI Database Assistant returns an interactive widget in the chat panel. These widgets do more than display information. They help you explore the returned data, refine the results, and, for supported widgets, perform follow-up actions directly from the widget.

Oracle AI Database Assistant returns the widget that best matches the question that you asked. If multiple widgets can answer a similar question, Oracle AI Database Assistant attempts to select the most relevant widget.

Depending on the widget, you can use features such as:

View rich visualizations that help you understand the returned information at a glance.
Use the widget options menu () on the widget to export the displayed data to CSV, refresh the data, or maximize the widget.
Click the add filters icon () on the widget to add filters, display the selected filter labels above the widget, and use the applied filters to refine and narrow the results displayed in the widget.
Perform supported actions directly from the widget data. For example, in the Incidents - Table widget, which is a response to the predefined question: Show list of open incidents, you can acknowledge an incident, assign ownership, add a comment, suppress an incident, or edit the incident summary.