Creating an Endpoint in Generative AI Agents
-
On the Agents list page, select the name of the agent that you want to create an endpoint for.
If you need help finding the list page, see Listing Agents
- Under Resources, select Endpoints.
- Select Create endpoint.
-
Enter the following values:
- Name: A name that starts with a letter or underscore, followed by letters, numbers, hyphens, or underscores. The length can be from 1 to 255 characters.
- Description: An optional description
- Enable session: Enable this feature to keep the context of the chat session. You can't change this option later.
- If you enable session, specify the idle timeout in seconds. The default idle timeout is 3600 seconds (1 hour). This means that after the lack of activity between the user and agent for 1 hour, the session automatically ends, and the following conversations don't retain the context of the previous conversation. You can set it as low as 1 hour and as high as 7 days.
-
Enable storage of large output results: Enable this feature to store large responses from the agent in Object Storage. This feature applies only to the SQL tool.
Note
When SQL execution is enabled on a SQL tool and you enable this feature, Generative AI Agents stores the output in a.csvfile only if the SQL command returns more than 100 rows. Generative AI Agents doesn't store SQL results of less than 100 rows. -
If you enable storage of large output results, specify the following Object Storage details for storing the output:
- bucket
- Object prefix, if you have folders
- Retention period in minutes. Enter a number up to 1,440 which equals to 24 hours. The default value is 360 minutes which is 6 hours.
-
Select whether to enable the following guardrails.
- Content moderation for input (user prompt), output (generated response), or both:
- Disable: Don't apply content moderation.
- Block: Help identify and apply content moderation.
- Inform: Don't apply content moderation, but aim to inform the user if the agent detects content that needs moderation.
- Prompt injection (PI) protection:
- Disable: Don't apply PI protection.
- Block: Help identify and protect against prompt injection.
- Inform: Don't apply PI protection, but aim to inform the user if the agent detects content that needs PI protection.
-
Personally identifiable information (PII) protection for input (user prompt), output (generated response), or both:
- Disable: Don't apply PII protection.
- Block: Help identify and protect PII.
- Inform: Don't apply PII protection, but aim to inform the user if the agent detects content that needs PII protection.
- Content moderation for input (user prompt), output (generated response), or both:
-
Select whether to enable the following options for the endpoint.
- Enable trace: Whether to track and display the conversation history, including both the original prompt and the generated response, during the chat conversation. If you don't enable this feature now, you can always add it later by editing the endpoint.
- Enable citation: Whether to display details about the source of information for each chat response. If you don't enable this feature now, you can always add it later by editing the endpoint.
- Enable multi-language: Whether to support chatting in the following languages: Arabic, English, French, German, Italian, Japanese, Portuguese, and Spanish. When disabled, English is the default language.
- Enable human in the loop: Whether human input is required such confirming an agent's generated response before the response can be used.
If you don't enable these features now, you can always add it later by editing the endpoint.
- (Optional) Select Show advanced options and add one or more tags to the endpoint. If you have permissions to create a resource, then you have permission to update its tags. If you need help, see Tags and Tag Namespace Concepts.
- Select Create.