Ending Sessions from a Chatbot¶

This page is a technical reference for the ways a chatbot can end a chat session programmatically. For how this fits into the overall session lifecycle, see Session Status.

You have three ways to end a session programmatically from within your chatbot. All three behave identically once triggered — the session moves to PENDING_REVIEW, the end time is recorded, and any configured conversation-end events fire.

The End Session tool¶

Add the End Session tool to your LLM node's tool list. The LLM can then choose to end the chat when it judges the conversation is over.

The tool description presented to the LLM is: "End the current chat session. This will mark the session as completed. New messages will result in a new session being created."

In that description, "completed" means the conversation is finished — the session moves to PENDING_REVIEW. It does not move directly to the COMPLETE status, which only happens once the participant submits the post-conversation review.

The session ends after the chatbot's reply is delivered to the participant.

For full configuration details, see the End Session tool reference.

Use this approach when the decision to end the conversation belongs to the LLM — for example, "end the session once the user confirms they are done".

The end_session() helper in a Python node¶

Inside a Python pipeline node, the runtime exposes an end_session() helper:

def main(input, **kwargs):
    if should_finish(input):
        end_session()
    return "Goodbye!"

Calling end_session() ends the session after the pipeline finishes and the response is delivered. The returned message is still sent to the participant first.

Use this approach when the decision to end the conversation belongs to your custom logic — for example, a state-machine progression or a specific sentinel input from the participant.

Events with an "End the conversation" action¶

Configure an event whose action is End the conversation:

• Static triggers — fire on a lifecycle event such as a new chatbot message, a participant joining, or a conversation starting. Useful when you want the session to end as soon as the chatbot sends a specific goodbye message.
• Timeout triggers — fire after a period of inactivity. Useful for "end the session if the participant is silent for 30 minutes".

Use this approach when the decision to end the conversation should be driven by lifecycle conditions outside the pipeline itself.