Skip to content

Agent Configuration - Other settings

In this section, we will

Post-processing

Conversation closure interval

This parameter controls time period after which conversations with the Agent are automatically closed and post-processed. Default value is 1 day, but intervals of 3 days and 1 week are also available. Conversation closure and post-processing are handled by a regular job run.

Post-processing

Enable post-processing to collect metrics about conversations, like user sentiment, language, or resolution status. Metrics are generated with the help of a post-processing Prompt Template, for example DEFAULT_AGENT_POST_PROCESSING.

Post-processing is done on the conversation level. Message post-processing is not yet released.

UI Settings

Welcome message

A field to provide a default welcome message for the end-users. This message remains on UI and does not become a part of Agent's conversation (i.e. you will not be able to see it in Traces etc).

User feedback

Enable user feedback to collect likes and dislikes. For negative feedback, a comment can also be provided, so that Agent owners can analyze feedback. User likes and dislikes are included in Usage reports on both message and conversation level.

Sample questions

Provide up to 3 sample questions to help users start their first conversation with the Agent.

Channels

Please check the Channels page for detailed information on how to make your Agent available via various channels like web or messengers.

Conversations

Under the Conversations tab, history of latest conversations with the Agent is stored. Please navigate to Usage or Traces for a comprehensive overview of Agent calls.

Notes

Use this tab to store your notes and test inputs for the Agent.

Test sets

Specific Test sets for evaluating Agents are not yet available, but you can choose a RAG or Prompt Template Test set here and use it for quick testing of specific Agent actions.

To use this feature, select a Test Set from the dropdown and just click on a Test Set record to copy it into the user message field in Agent.