We have a questionnaire use-case where we're using tool calls to retrieve questions from a backend (to present to the user), and then async tool calls to persist their answers. The conversational experience is pretty poor however, since the agent has to (repeatedly, for each question) announce that it is starting a tool call, announce the success of the tool call, pause and wait for the user to say something, then ask the question (the behavior defined in the prompt), get the response, and announce the success of the persistence call, and again wait for the user to say something before it proceeds. We would really like to A) suppress tool call start and success messages, and B) eliminate the need for a user interaction before proceeding with the prompted tasks. This is an example of the task portion of a prompt which still exhibits all of the above problems:

[Task]
1. Get a question for the user by Invoking the 'get_next_question' toolCall with the parameter 'XXXXXX'.....
  - immediately pose the returned question to the user
  - wait for their response
2. Send the response to the server by invoking the 'persist_user_response' toolCall with the 'XXXXXXX', 'YYYYYYYY', and 'userResponse' parameters.
  - do not wait for a response from the user
3. Repeat the response back to the user and politely end the call.

Is there a way to silence the status messages for both synchronous and async tool calls? Leaving the configured message empty just results in some default message. Is there a way to allow the agent to proceed without user interaction after a successful tool call? For simplicity and determinism, we would really prefer to keep these granular calls rather than returning all the questions in one response and then aggregating all responses and submitting them together. Thanks!

Heymark0102038710\_44828 Thank you for creating the ticket! A member of our team will reach out to you shortly.

@mark0102038710_44828 I understand your concerns about the conversational flow when using tool calls. To address your needs: A) To suppress tool call start and success messages, you would typically need to adjust the settings in the backend or the API that is responsible for the voice interactions. This might involve modifying the verbosity settings or the interaction model to prevent the system from announcing the status of tool calls. If such settings are not directly accessible or modifiable, you may need to work with the backend development team to implement these changes. B) To eliminate the need for user interaction before proceeding with prompted tasks, you can adjust the conversational design to move seamlessly from one step to the next without explicit confirmation from the user. This would involve designing the conversation flow to assume success and continue unless an error is encountered. For both of these adjustments, if the current system does not support these configurations, you may need to reach out to the support or development team responsible for the voice API or the backend system to request these features or to inquire about potential workarounds. If you have the ability to modify the system or if there are existing settings that can be changed to achieve these results, I would recommend exploring those options. Otherwise, please ask support for more info on this, as they may have additional insights or solutions that are specific to the system you are using.

Hey @mark0102038710_44828 To help track down this issue, could you share: - The call ID - What response you expected to get - What response you actually got instead This would really help me figure out what went wrong!