How can i start a call with the client SDK using workflows instead of an assistant?

from webhook received for calls from vapi how can i decide that call successfully completes or failed so that i can retry call

What's the difference between HIPAA compliance on the ADD-ON page and HIPAA compliance on the assistant's privacy settings? I see that on the add on page it's +$1000 but I can still enable it for the assistant in the privacy section. What's the difference?

I want to be able to identify user from cookies token to make sure I'm only retrieving data for a particular user when custom tool is called

Is there a way to place a SIP trunk call without explicitly setting the caller ID (such as through the P-Asserted-Identity or From header), and instead let the SIP trunk provider automatically apply the default caller ID assigned to the trunk?

Every workflow I build I get the same error https://cdn.discordapp.com/attachments/1380823910597267558/1380823911284998194/Bildschirmfoto_2025-06-07_um_10.17.13.png?ex=684547fa&is=6843f67a&hm=1b77e7c38a60cda80999116b6e9ca7b81bfc5a3dc0c3ae754c0b4a9c0a3a48a3&

we have requirement when plan is expired or credit exhausted, all incoming call should be hangup so as of now there is not way to support this.

We're migrating from Twilio to Vapi. Currently, our customers can enable "voicemail mode" where the agent speaks once and then the caller leaves a message and hangs up. This is a nice bail out when their agent is not answering correctly, etc. With our Vapi assistant, i tried to set "waitSeconds" to a large number but it errors out > 5: startSpeakingPlan: { waitSeconds: 180 }, ... After testing a bit, these "hacks" work: - Having the llm reply with a space - Delaying the LLM reply I'm just wondering if there's any recommended approach. Thanks

i sent info in on the website but havent been reached out to yet!

I have tried to create a project, where the user makes a call to the web assistant. However, when I am trying to make a call, the assistant is not talking in the logic of workflow, only as a regular call to AI ("Hello, how can I help you today" and etc). I am using the assistant id and have putted it into my .env file, which i am proceeding further in code to make a call. When I am trying to put the workflow id instead, the issue of "Couldn't Get Assistant.

assistantId

c84c370e-880b-4b87-b052-b4477c896703 Does Not Exist." appears. The assistant id that I was using is: 79f3c469-97a4-4c3c-882b-bab3302465e4 , and the workflow by which logic assistant had to speak is: c84c370e-880b-4b87-b052-b4477c896703

Agent Idle message saying "Oh my god are you still there?" when set to "Are you still there?" Call ID: ed0619da-08be-42bd-ae40-5dbea6566549 Timestamp: 11:53

workflow ID: a43e4d98-fe6b-4f7e-b5d7-49ef72675053 1. Tried GPT 4.0, 4.1, 4.1 mini, 40 cluster no effect 2. Added & removed stop node after transfer call no effect. 3. Tried single prompt agent no workflow, call transfers still fail 4. Selected all the warm call transfer options/modes, and blind transfer with no effect 5. Added removed Enforce E164 format 6. Ensured the Phone Number is good its just my personal cell phone 7. Request Start Message Default, None, Custom no effect 8. Removed all spoken messages during for any and all stages of tool execution

I have a cold call agent with a very good prompt but when i added my call transfer tool to the assistant its all gone out the window. Does anyone have a good idea of how i can fix this?

I want to make outbound call and if agent have some information about user than that would be great.

how to specify to use my sip trunk

The tool "send text" doesn't have a place to customize what text is being sent

I tried connecting number from twillo to Vapi, and every time i call it im getting Application error, and i see this in Twillo logs: Message Got HTTP 404 response to https://api.vapi.ai/twilio/voice Error Description We try to provide specific webhook errors whenever possible, however in this instance we are unable to identify the exact cause of the webhook delivery/response issue. To prevent 11200 errors, ensure that your webhook receiving infrastructure can quickly respond to inbound Twilio requests with a 2xx status code.

Currently I am trying out the workflows feature. I created a basic workflow but for some reason the web call isn't working, Please could you help?

Hello I am getting inconsistent webhook errors both on Assistant Request, Status Update, and End Of Call Report. They are 500 errors, and are responding with timeouts of 7500 ms and 20000 ms. In the past the call.in-progress.error-vapifault-custom-llm-llm-failed error has been because our service was taking too long to return the llm response, but after digging into our logs I am seeing that these invocations aren't being made on to our service. My assumption is there is something going wrong with VAPI calling our endpoint. On top of this I've tried to replicate the issue by calling our endpoints via postman concurrenlty and received no timeouts. I've listed a few call ids and and full example with our logs below. It seemed to start around 4 pm yesterday. This is very urgent as we are losing business. Please any assistance would be great. Call ids: 96400d2d-338f-4178-a95d-f4444e326328 621336e3-249d-4c7a-a30d-16b5644d53d3 @Shubham Bajaj

can you please provide me detailed create assistant post request which all variables it gets , as i want to conduct surveys using assistant , i want all variables to be sent in assiatant as in sdk system prompt is not available

i want to implement surveys for my ehr application how to do it using vapi , best way to do this please let me know i have 50 patients and i want to sent survey asking some things

how can i enable keypad inputs in workflows , basically i am building an workflow of creating surveys in which ai will asks questions and record inputs how to implement this any suggestions

I have enabled them but on IVR it is not working and call is getting ended

[
  {
    "id": "e16e22e7-08c4-45a1-a889-41cbf0b36f1e",
    "orgId": "01fc90d6-f3d4-425b-bd2e-5d35728fde87",
    "sessionId": "613c3bbf-aebe-4c2f-a976-37787a559267",
    "input": [
      {
        "role": "user",
        "content": "Get availability again"
      }
    ],
    "output": [
      {
        "role": "assistant",
        "tool_calls": [
          {
            "id": "call_iQQRD0wPlIbAvPhzECqV9K6V",
            "type": "function",
            "function": {
              "name": "get_available_slots",
              "arguments": "{\n  \"region\": \"colorado\",\n  \"leadaddress\": \"2091 East 74th Avenue Unit C, Denver Colorado 80229\",\n  \"leadpreference\": \"night\",\n  \"requires_onsite\": \"No\",\n  \"quote_or_estimate_only\": \"No\",\n  \"leadTrade\": \"solar\",\n  \"leadTimeZone\": \"America/Denver\"\n}"
            }
          }
        ]
      },
      {
        "role": "tool",
        "content": "No result returned.",
        "tool_call_id": "call_iQQRD0wPlIbAvPhzECqV9K6V"
      },
      {
        "role": "assistant",
        "content": "It seems I'm still not seeing available times at the moment. Would you prefer to pick a day and time that’s convenient for you, and I’ll try to reserve it for your virtual solar consultation?"
      }
    ],

Keeps returning no result returned but its never hitting my webhook url.

there are many many mismatch bw type in vapi sdk typescript , cerating workflows , assistants using sdk many options are not present in sdk available 0.9.1 , like in workflow messageplan is not available also in assiatents system prompt not available also it not matches with doc provided on website please help me i am stucked in creating workflows using apis using sdk #1211483291191083018 @NicxOp1 using dashboard i can do it which i can not do using apis either sdk is not updated can you please update it

I’m planning a political campaign where we need to make approximately 150,000 outbound voice calls (each lasting less than one minute) over a period of 5 hours. We are considering purchasing 490 additional concurrent lines to support this load. We will be using a landline phone number from Vonage. Before proceeding, I’d like to confirm: Is there anything else we need to consider or configure to ensure smooth execution of a campaign of this scale? Are there any limitations, rate limits, or restrictions we should be aware of? Do you recommend any specific best practices or configurations for handling this volume efficiently? Thanks

We are testing a demo of a personalized conversational bot that should use the historia_personal.json file to respond based on real user data. However, the bot is not accessing the file, and is responding with generic or invented content not present in the data. @Sahil assistent id: df57a123-88bd-422e-a8fc-4ef0a33052c9 https://cdn.discordapp.com/attachments/1380582628771758111/1380583123770933299/image.png?ex=684467ba&is=6843163a&hm=1fc1f06cf5999fdb9f47b75c65452e05b4fadd07ffdca3bb79a12acf33d2e03a& https://cdn.discordapp.com/attachments/1380582628771758111/1380583124248821780/image.png?ex=684467ba&is=6843163a&hm=7313c70f33d506432f931c98c22a7fe80738708a7bce3049fb5ca7af19926adf&

How do you integrate Vapi with a Tmobile business number?

Hey Vapi team — I’m trying to start a call using the /call endpoint from n8n, and I keep hitting this error: 400 - "Couldn't Get Assistant. Need Either

assistant

,

assistantId

,

squad

, Or

squadId

." Here’s what I’ve tried so far: Passed assistantId: Also tried using "assistant": "Kylie" by name Tried both a Twilio number (with SID and auth token) and the Vapi shared number using phoneNumberId: Using n8n’s HTTP Request node to send the request Set the Authorization header with both Private API Key and Public API Key Also tried authenticating with a JWT (including orgId, token, and exp) in the Authorization header Still getting the same error on every attempt I’m currently on the trial plan Can someone confirm if trial assistants need to be saved, assigned to a squad, or have special permissions to be used via backend API? Appreciate any help — I’ve been troubleshooting every angle. Thanks

I really liked Vapi's Workflow, but I'm trying to use the workflow WebSocket connection for my use case and haven't been able to find how to do that.