Hello I've been struggling with a problem and can't find a solution, so I decided to finally contact you. I don't know how to send information back to vapi during a call. I've created a voice agent, which retrieves a question from the user and sends it to make via webhook. A perplexity module executes the instruction and I've placed a response webhook to send the answer back to vapi. Unfortunately I don't know what information I need to provide so that the information is returned to the right vapi agent, on the right call. Do I need authentication? An api? A call id? In short, please guide me. Thanks in advance

Hi When using a webhook you also need to include a webhook response, if this is an function you will need to format it as such with the toolCallid, and data you want to send to the agent as the result. { "results": [ { "toolCallId": "call_VaJOd8ZeZgWCEHDYomyCPfwV", "result": "San Francisco's weather today is 62°C, partly cloudy." } ] } https://cdn.discordapp.com/attachments/1249779895761109022/1331867649382027274/image.png?ex=67932de8&is=6791dc68&hm=74e432d2fffebba8c19bb798ea0ce9476d7bf5945c95106d8c5636043f6acc5a&

Thank you very much for your detailed reply. I tried but same result. Doesnt work. From webhook response, i can find input but no output. From input i can see call id and my perplexity request result, but vapi can't receive it at all. I tried from vapi to create new properties to receive webhook reply, dont works too. 😔 https://cdn.discordapp.com/attachments/1249779895761109022/1331909293502365716/chrome-capture-2025-1-23.png?ex=679354b1&is=67920331&hm=2048ba9b47d71a87a2d767d7cefc59abe77b8257611aa2b0980ba7065caf8d65&

Just tried by vapi respond to a function call module, same result, model can't receive and read results during call https://cdn.discordapp.com/attachments/1249779895761109022/1331911879106363424/chrome-capture-2025-1-23_1.png?ex=67935719&is=67920599&hm=976a6d8cb68765db4cdc63c610d8bbee1b8a8af30058b098f407141b662c0d03&

I am encountering the same issue. If someone could help that would be awesome !

When you are responding back to vapi it needs to be a proper json structure.. Mostly there are problems with spacing... It's technical.. you can use chatgpt to properly structure your response first,, then send it back to vapi... with the webhook response....

Has it never worked? I’m using Make webhook response node with the same json content and it does work for me most of the time. At times after successfully running the tool the call ends abruptly. Otherwise it works. Also make sure that the content you’re sending back doesn’t break json formatting

Not worked for me, double checked json with chatgpt 😢

Looks like the new lines are breaking your json. Maybe add another openai node after perplexity (or use any ai node or code node) and ask to give the output with escape characters so that it doesn’t break jaon formatting. This should ideally solve your issue i hope

Thank you for help, i will just remove my variables and add one sentence to test and let you know

i can assure you that is the issue.. it happened with me when i was starting out... i spent hours just to find out it was a new line not properly escaped..

Thank you @jithins and @zehaan for your help. 1. I can confirm its related to content maybe breaking json. I removed perplexity output content variables and added one sentence as result. I can hear vapi voice agent read "Ok, done". Its my setting when tool Request is Completed (Message shown when the tool completes successfully.) But it doesnt read my one sentence set as result. Not perfect but i think its interesting. 2. I removed my one sentence result and placed "transform to json" module before sending his output to vapi, bad result again. Voice agent saying "no result to read" as it did in the past 😢 You can check screenshot. Thanks to all https://cdn.discordapp.com/attachments/1249779895761109022/1332394169532153867/chrome-capture-2025-1-24.png?ex=67951844&is=6793c6c4&hm=1caafb2b2e9434b6f260dfc8411842e838e007868b332a30ee0ad92d4e88c7f3&

Is tool set to be async ?

no async, its not enabled. You can also check my tool setting and prompt in screenshot https://cdn.discordapp.com/attachments/1249779895761109022/1332463705224646687/async_and_properties.png?ex=67955907&is=67940787&hm=f81582feb5d052130b02a7a85674b7002e62e4cc076dc9b56d27414d6b37eebd&

sent you a message

why you all using Make instead of N8N

@Armel The module before that needs to be CREATE Json .... In any case you need to share all the input/output bundles if you want proper help. We can't diagnose those screenshots... FYI use Claude/Deepseek R1 , will easily get it solves & you can share the full JSON automation file with them...

Ideally you would not need to do that and jsut paste the JSON, but nice.

Anyone has an idea where to get a turkish number for voice agent?

Hay forma de hacer que la AI califique clientes haciéndole preguntas potenciales y si son positivas pasarlo de una etapa a otra en GoHighLevel?

message has been deleted

Who needs help with a vapi project

Hello Can someone explain how I can create my own vapi button instead of using the widget? I want to use an animated gif that triggers the call. But my big problem is that the vapi widget is really difficult to place where I want on the page, apart from the predefined choices planned top left, top right, bottom left, bottom right, I'd be delighted to find a method where I create my own button and place it wherever I want. Thank you

Tengo un problemita, no me quieren aparecer en MAKE las etiquetas del webHook, estoy configurando un escenario con Vapi para agendar reunión en GHL pero las etiquetas para que tome los datos no aparecen, ¿Alguna ayuda?

Hey

I want to know about that can two vapi agents call the ghl calendar simultaneously ?

What’s everyone using for text/email ai with GHL?

To create a custom Vapi button using an animated GIF, you can use the Vapi Json API to trigger the call. This allows you to design and place the button anywhere on your page. Simply embed the GIF in your HTML, add an event listener to trigger the call, and position it as desired. Let me know if you need further guidance