Issue with Tool usage
# support
v
I am trying to create an Appointment setter using Vapi and Make My assistant uses 2 tools named Check and Book to check the availability and book an appointment My assistant queries the webhook configured on the tools after gathering the information and Makes automation complete without any issues The Assistant isn't saying anything or behaving according to the response received It just says, one moment, still checking.. etc but not getting the response I need and is there a way to test the tools like we used to for functions #1245660177953194066 call ID 475619cc-2dda-47bb-960e-65d41d375c88 call ID 45bc6a56-a07c-41e2-a190-b3374339e930
v
@Vj It sounds like you're experiencing an issue where your assistant is not providing the expected responses after invoking the Check and Book tools. To address this, we need to ensure that the assistant is correctly processing the responses from these tools and then communicating the results back to you. Firstly, let's confirm that the assistant is correctly configured to receive and interpret the responses from the Check and Book tools. The assistant should be designed to handle the JSON response format you've described and then provide a meaningful reply based on the "result" field. If the assistant is stuck saying "one moment, still checking..." it could indicate that it's not receiving the response, or there's an issue with the logic that handles the response. To troubleshoot this, you can perform the following steps: 1. Check the server logs to ensure that the HTTP requests from the Check and Book tools are being received. 2. Verify that the server is sending back the correct JSON response format. 3. Review the assistant's code to confirm that it's designed to parse the JSON response and act accordingly. 4. Test the individual tools (Check and Book) by manually sending HTTP requests to the webhook URLs to see if they return the expected responses. Unfortunately, there isn't a built-in test feature for tools like there is for functions, but manual testing using HTTP requests can serve as an alternative. If you've gone through these steps and the issue persists, please provide more specific details about the assistant's behavior and the responses you're receiving. This will help in diagnosing the problem further. If you need assistance with the server-side implementation or the assistant's code, please ask support for more info on this.
You can continue this conversation by mentioning me in the message.
v
Here is my webhook response Body { "results": [ { "toolCallId": "0361a3b8-5c03-4aaa-babb-fc2b78bfb492", "result": "All time slots are available between 2024-07-22T09:00:00.000000+05:30 and 2024-07-22T17:00:00.000000+05:30." } ] } Status 200
Another:: { "results": [ { "toolCallId": "call Id", "result": "We are not available on weekends. Our office hours are Monday to Friday 9am to 5pm." } ] }
s
@Vj 1. your returning the wrong tool call id, the tool call is something like this "call_ya4zepYiLuybPozRO0Wydu0O" 2. make sure your tool response is correct because it's coming as plain string please double check your make workflow https://docs.vapi.ai/tools-calling#server-response-format-providing-results-and-context
v
is there a way to test the tool without calling the agent with in Vapi Dashboard
I am ending up using more credits .. while testing the call flow
Is there any way to test tools without actually speaking to the assistant
g
Made a new video on how to use Tools instead of functioning for making appointments setter agents with Google calendar.

https://youtu.be/n4Ii0efZ8yM

s
@Vj is your issue resolved now?
2 Views