Hi team, this Assistant records details from the customer and sends this info in an email to the manager. With this specific call, the customer provided their Phone Number and the Assistant confirmed it back __correctly__. Although, the Data in the email shows the incorrect phone number (screenshot attached). Any insight on how this happened? Call ID: dd5b2233-e956-4df4-9a10-559fec7b0fe7 https://cdn.discordapp.com/attachments/1351424752073314334/1351424752262184962/image.png?ex=67da53e1&is=67d90261&hm=4f7701d2896f8d6707b71b1628d61d8a0ccf3e7fa28928f7be4a1cb6a825c974&

Adding an explicit warning section to the system prompt would help ensure the AI doesn't modify or "fix" user inputs with assumptions or placeholders.

[Warning]
Do not modify or attempt to correct user input parameters or user input. Pass them directly into the function or tool as given.

This would be particularly effective for several reasons: 1. It creates a clear instruction that parameters should only be populated with actual user-provided information 2. It prevents the AI from inserting placeholder or "default" values when data is missing 3. It would force the AI to properly collect all required information before making API calls I would also recommend adding a directive to have the AI explicitly verify when a required field is missing:

If a required parameter is missing or unclear in the user input, explicitly ask the user for that information before proceeding with any function calls.

This way, the assistant would be required to postpone the API call until Jenny had actually provided her phone number, rather than proceeding with an assumed or default value.