### Answer
This is a known timing issue when using
Skip Response on a Conversation Node followed by a Function Node. [(1) ](
) , [(2) ](
)
When
Skip Response is enabled, the transition happens once the agent finishes speaking, but the execution can race ahead to the Function Node before the speech completes audibly. [(1) ](
)
## Workarounds to try:
1. **Use "Speak During Execution" on the Function Node instead**: Enable
Speak During Execution on your Function Node and move your static text there. This ensures the agent speaks while the function runs. [(2) ](
)
2. **Disable "Wait for Result" on the Function Node**: If you don't need the function result immediately, turning off
Wait for Result allows the transition to happen once the agent finishes speaking (if Speak During Execution is on) or immediately after the function is invoked.
3. **Add an additional Conversation Node before the Function Node**: Some users have worked around this by adding a separate Conversation Node with the desired speech, then transitioning to the Function Node. This forces the speech to complete before the function executes. [(3) ](
)
4. **Use a faster model**: Some users have reported that using a faster model (like GPT-4.1 Fast) and adjusting the Speak During Execution settings can help with timing issues. [(4) ](
)
The core issue is that the Conversation Node's static text and the Function Node execution can overlap in timing. Moving the "filler speech" to the Function Node's *