Conversations and tasks

Users interact with agents through conversations for business Q&A, content processing, and task execution. What an agent can do depends on its configured knowledge bases, skills, and tools.

Ways to use an agent

Usage

Description

Q&A

Ask the agent a question to get information, an explanation, or a recommendation

Task

Submit a clear objective for the agent to process and return a result

Ongoing conversation

Ask follow-up questions, provide more information, or adjust requirements based on an existing result

  • With a knowledge base, an agent can answer questions using business materials.

  • With skills, an agent can follow defined methods for specialized tasks.

  • With tools, an agent can query external platforms or perform actions.

Because agents have different knowledge bases, skills, and tools, they are suited to different tasks.

Use an agent

  1. Open the Agents page.

  2. Select the agent to use.

  3. Start a new conversation.

  4. Enter a question or task request.

  5. Upload relevant files as needed.

  6. Send the request and wait for the agent to process it.

When submitting a task, describe the objective, the content to process, and the expected result format.

Process a task

After receiving a task, the agent selects relevant knowledge bases, skills, or tools based on the task. During processing, you can:

  • Provide information requested by the agent.

  • Confirm actions that require user authorization.

  • View the agent’s response.

  • View and download files generated by the agent.

  • Ask for changes in the current conversation.

You can stop the current task if you do not want it to continue. Stopping a task does not undo external actions that have already completed.

Continue a conversation

You can continue asking questions, add information, or revise task requirements in the current conversation. You can also start a new conversation for another task. Conversation history lets you review previous questions, tasks, and results.