I have a function tool and if I do context.wait_for_playout() in it then what will happen

@function_tool
    @log_exceptions(logger=logger)
    async def provide_club_information(
        self,
        rephrased_user_query: str,
        context: RunContext,
    ):
        """Retrieve club-related information or policies based on the user's intent.
        """
        logger.debug(
            f"User Query Rephrased by provide_club_information tool: {rephrased_user_query}"
        )
        await context.wait_for_playout()

because I am getting error like this RuntimeError: cannot call

SpeechHandle.wait_for_playout()

from inside the function tool

provide_club_information

that owns this SpeechHandle. This creates a circular wait: the speech handle is waiting for the function tool to complete, while the function tool is simultaneously waiting for the speech handle.

Thanks for reaching out, Shataxi. I'm an AI assistant and I'm here to help. Give me a moment while I research this topic.

But we do ctx.wait_for_playout() in end_call tool and it works perfectly there then what is the problem in my case

Thanks — that’s a good observation. The difference comes down to when the tool runs in the speech lifecycle and whether the tool itself is part of the response that is being played out. Why it works in your end_call tool • The

end_call

tool is typically invoked after the agent has finished generating the response and the speech playback has either finished or is in a state where waiting for playout does not depend on the tool completing. In other words,

end_call

runs outside (or after) the active speech’s critical path, so awaiting

ctx.wait_for_playout()

there does not create a circular dependency. Why it fails in provide_club_information •

provide_club_information

is being executed as part of the agent’s response generation. The agent will wait for that tool to return before it finalizes the response and completes the speech lifecycle. • Calling

await context.wait_for_playout()

inside that same tool makes the tool wait for speech to finish, while the speech system is waiting for the tool to finish. That is a circular wait — the speech handle cannot complete because the agent hasn't finished (the tool is still running), and the tool cannot finish because it’s waiting for speech to complete. That deadlock is what raises the RuntimeError. Key difference summary • end_call: runs after/outside the active speech path → waiting for playout is safe. • provide_club_information: runs inside the active speech path → waiting for playout causes circular wait and is disallowed. What to do instead • Move

await context.wait_for_playout()

out of

provide_club_information

into the caller/agent flow that invokes the tool, or into a tool that runs after the response/speech completes (like

end_call

). • If you need follow-up work when playback finishes, attach that work to a post-playout handler or schedule it as a separate async task that is triggered after the agent’s response completes. • If you share what behavior you want after playback (e.g., log analytics, start another tool, send a follow-up message), I can suggest a concrete pattern or code example that avoids the circular wait. If you want, paste the call site where

provide_club_information

is invoked or describe the exact sequence you expect after speech finishes, and I’ll propose a refactor you can use.