Does anyone know why my bot is saying
# archive-prompt-sharing
p
Does anyone know why my bot is saying sorry right before hanging up? I'm using make.com to book appointments. Instead of returning the function call (because that wasn't hanging up correctly either) I'm injecting speech directly into the call the control url. The last thing it SHOULD say is "I'm sure we'll chat soon." THEN HANG UP. But it's saying "I'm sure we'll chat soon. Sorry" then hangs up. Usually the last syllable of sorry is also cut off. https://cdn.discordapp.com/attachments/1246367057046405130/1350639809936359496/image.png?ex=67d778d8&is=67d62758&hm=a2374837dc8298944861092fe1e6ab39b3b7cf3821b2febf696f38c6229bac13&
This is the final piece of the prompt: ## 4. Booking Process - Once the prospect picks a day and time, run the function
Booking-Royce-CGP
including their name and their requested booking time. - If they ask for another time, acknowledge their preference and run the function with the updated slot—ensuring it meets the booking constraints. - If the booking was successful, do not do or say anything. They will hear a confirmation message, and the call will be terminated.
Also interestingly it's not showing up on the transcript
I seem to have fixed it by changing the model from gpt 4o mini cluster to 4o cluster and returning a response to the function call 'booking successful' Then at the bottom of the prompt telling it to say the same thing and then hang up.
A little disappointed I can't get it working with 4o mini because the latency is noticably lower, will keep looking
t
this happened to me when I got a response from make (we use it as well) which is 500. 400 produces a "hang up" and 200 is working as expected. You should also try to make it asyncronous (there was a setting in the assistant or tooling page for it) so it doesn't wait for the response if possible
4 Views