I'm following through the guide at https://docs.vapi.ai/advanced/sip/telnyx. I'm now at the part where I have to configure the SIP invite, however, I'm unable to set my URI due to this error. https://cdn.discordapp.com/attachments/1391542583431401573/1391542588947038219/image.png?ex=686c4686&is=686af506&hm=ddc5791af30cad90f91104cb04ffcec817f675d69d9188d2a6e6c05ad3c3c0a8&

I'm not able to start the call, I have active credits and jsut get an error message, not much details of what exactly went wrong. here is the call ID: ec2b7989-2cea-41a8-95a8-7395655a6495 https://cdn.discordapp.com/attachments/1391506302261264494/1391506302873895042/image.png?ex=686c24bb&is=686ad33b&hm=853c32c7a998fdd44bfdda95e846fd1db06fdc79a8f67270b7c997623595568d&

Hi Vapi.ai Support Team, I'm encountering an issue with one of my voice agents related to email address recognition and confirmation. In my current script, I’ve added a part where the voice agent asks the caller for their email address and then confirms it by spelling it back to the user. The caller also spells out their email address when providing it. However, the voice agent is having trouble correctly understanding the email address and spelling it back accurately for confirmation. Details: Issue: The voice agent misinterprets or incorrectly confirms the email address, even when the caller spells it out clearly. Expected behavior: The agent should accurately recognize the spelled-out email and repeat it correctly for confirmation. Script behavior: The agent prompts for the email, the user spells it, and the agent attempts confirmation but fails to capture or repeat the correct address. Could you please help me troubleshoot or optimize the script/agent settings to improve email recognition and confirmation accuracy? Thanks in advance!

While Designing Wrokflows the dashboard stopped working @Kings_big💫 https://cdn.discordapp.com/attachments/1391465143895855185/1391465144663543989/image.png?ex=686bfe66&is=686aace6&hm=40281c9c72aa935b83efb4398d649589ac1fbf95343c2a2e6b0b6ad54471d826&

Hello I have recently created a new conversational agent in workflows although we are experiencing a ton of issues. The main issue is that the assistant randomly starts repeating the same message within the conversational node and ends up breaking itself. Example: Call ID: aff34aec-6d72-4d54-830f-7ddf5cf09b38 Timestamp: 1:25 We are also noticing issues with transcription. Are there any enhanced English transcription settings we could use? We are currently using: Deepgram Nova 2

There is no provider keys option option in my dashboard i want to use google calendar inbuilt feature https://cdn.discordapp.com/attachments/1391349157783932928/1391349158157353000/IMG_3362.jpg?ex=686b9261&is=686a40e1&hm=c70a1836f96395117e3efb6aa65d1ca019c8486f733126e4a91ff05b10400ec2&

why is vapi giving incorrect times? when using "{"now" | date: "%A, %B %d, %Y, %I:%M %p", "America/New_York"}

I am using Vapi workflow and for backend. I am using python with fastapi framework. So right now what I’m doing is my agent is taking data from Google sheet like there is name phone number and email so when a call initiate from that number, it can recognise the name and email and send the email to that person. That is what my backend is handling and I am passing webhook to schedule meeting to apirequest tool to Vapi workflow to implement that email sending, but what I want is Vapi to greet my customers with their name, just by recognising the call like the call is coming from this number and recognise the customer number and greet him with his name. How can we do that with vapi workflow?

Hi, I am new to Vapi and building a fitness ai agent while following a tutorial on the youtube. But the tutorial has used the old version of Vapi, so by following along I can't connect my assistant to workflow it should go about. Can anyone help me set it up. I have to complete this proj asap for college resume

How can we use a query tool effectively in workflow because right now I’m using a query tool which have my knowledge base and my work flow is not triggering query tool to answer the questions. How to resolve this. https://cdn.discordapp.com/attachments/1391113688643146000/1391113689016565830/IMG_3355.jpg?ex=686ab715&is=68696595&hm=763c0f6d4391b2a1b0e4e7337e750a2f4584c16754dfc594481594d48d6d6a89&

Hello, I have been trying to add more credits to my account and I cannot login anymore. After successfully submitting all the ID Verification data, the pop up remains loading indefinitely. The email associated with my account is luis@dynamocollective.com. Can you please advise? Thank you

Is it possible to use voicemail detection with workflows and how? i was searching in the docs but couldnt find anything

Hi there! Have a question: Is there any possible way to save the phone number of the person who calls on my assistant? I don't want to make the assistant to ask the phone number

Im following this guide https://docs.vapi.ai/advanced/sip However after creating a sip phone number its not possible to call my agent, i have tried exactly as the guide says using linphone

I can't find anything in documentation.

If you goto dashboard, and click on assistants, you get the right panel with the main toolbar at the top If you click on any of the menu options (example tools), the top panel drops off. it is thus not possible to make any edits, or save any changes via dashboard. https://cdn.discordapp.com/attachments/1391029838655258634/1391029838906920960/image.png?ex=686a68fd&is=6869177d&hm=f9c5eb77d9149be5aa3968d1fdab1580ed7950258c16973808dc4d048286201e&

Hi. I'm having trouble with workflow publishing. I created my VAPI account about 2 weeks ago, and I'm trying to use workflows. But I only see a save button without any publish option. Could you help me get access to the full workflow features? My workflow is saved, but I can't activate it for use in my project

When trying to create an MCP tool via the API there isn’t a key to set a name for the tool. The VAPI api doesn’t accept a name field to name the tool like the other tool calls (eg API Request )

Hello. Are calls via zadarma trunk can be forwarded to another number (external) or SIP with SIP refer only? In your instruction guides I've found this method: SIP REFER (call transfer) If you need to transfer a call to another number, you will need to add a SIP Transfer based call forwarding where the transfer number will look like this: sip:transfer-number@your-telecom-provider-domain.com Example: sip:15557654321@sip.zadarma.com Unfortunately, it doesn't work, I've checked .pcap of calls during the transfer, the method is not supported (501), the calls are being dropped on zadarma side, and there is no option to activate the method within zadarma. protosartos — 00:13 Do you have any other solutions? Should I set the transfer somehow differently? The objective: to forward incoming calls from AI assistant to a human (external number) or SIP authorized device

I am using workflow and doing batch calls from my backend, i added a api request tool just before the call end node so that my batch call knows that the call has been disconnected and call to next person, but what if a user disconnects the call early i am not able to handle that, can anyone please help with that ?

I've noticed that the google transcriber works fine on the assistant section, though when I use it on workflows it doesn't work. https://cdn.discordapp.com/attachments/1390786827149115422/1390786827341922374/image.png?ex=686986ab&is=6868352b&hm=eedeaceb04e517f852086b457cb996ba662d003a171978653f78249dce9b3a8d&

In the last 2 weeks I tried to connect a custom tts to a vapi agent. But it just doesn't seem to work properly. First of all, why do you need to create the custom tts only through an api call??? I think it would be much easier if this option was possible from the UI. Second, I tried to send api calls like this: curl -X POST https://api.vapi.ai/assistant -H "Authorization: Bearer my_api_key" -H "Content-Type: application/json" -d "{\"name\":\"Hebrew Gemini Agent\",\"model\":{\"provider\":\"openai\",\"model\":\"gpt-4o\"},\"voice\":{\"provider\":\"custom-voice\",\"server\":{\"url\":\"my_server_link\",\"secret\":\"mysecretkey\",\"timeoutSeconds\":30}},\"transcriber\":{\"provider\":\"deepgram\",\"model\":\"nova-2\"}}". Every time I use this, it creates an agent, but when I enter the vapi app, it shows this( the picture) https://cdn.discordapp.com/attachments/1390784364341624924/1390784364589092965/IMG-20250704-WA0037.jpg?ex=68698460&is=686832e0&hm=11238e18079d74384c62823479991fdf8ad54489b114f0e63efa11b96e2f58d8&

I am getting this error in my workflow: Workflow id: 29c71ef0-5a66-40f0-9760-44434bb61426 Please resolve this error @Vapi @Vapi Support Bot @VAPI Tech https://cdn.discordapp.com/attachments/1390772093171269643/1390772095494918315/IMG_3342.jpg?ex=686978f2&is=68682772&hm=81c5b7f001df957fd8d9f902da920365ee44274722021a10b09347276a6ca7e8&

Hey folks so I grabbed a fresh Telnyx number (HD quality, never used before, etc.), everything looks good, but I'm getting a solid 8-9 second delay on outbound calls from Vapi (delay of the agent to answer) Same exact setup (same assistant, tools, flow), but when I switch the number to a Twilio one, the delay drops to like 1 sec or less, super smooth! Interesting part is when I get the recording from Vapi, its like "stitched " perfectly without the delay gaps you can clearly experience on the phone call. The record sounds perfect, but it wasn't like that on the call.. Twilio:

c992400a-87e6-4cd0-9bc1-96f8da75f8ac

| time: 1751654238450 Telnyx:

963c3c94-7c26-4c8b-b710-6ef38ee72b05

| time: 1751654107674 From the recording it feels like the audio was sent but not received. Really want to get this working with Telnyx, anyone else run into this? is there a fix or a setting I’m missing? Appreciate any help 🙏

I created and scheduled a voice campaign via Vapi’s API using the Asia/Karachi timezone, but the campaign remains in “Scheduled” status and no calls are being executed.

How can I force tool call values to be forcibly set to null? Specifically, this use case lays inside a WF, and this tool can be called in multiple paths, so it is modified with different values set to null depending on the section of the flow the call is in. https://cdn.discordapp.com/attachments/1390726461270397059/1390726461861662810/image.png?ex=68694e72&is=6867fcf2&hm=8fe03c761d6443f85a6f5f6735a8239acf34d7b52648fe2bdebfec48bff3f1b3&

I have a workflow that on the first nodes works correctly, until it arrives to a node where the user has to say the plate number of his vehicle. The flow stops there I think as it does not detect correctly when the user has finished saying the plate or something like that. I have created a new workflow with two nodes, to test, and it keeps stopping when the agent asks for the plate. When it asks it and you say a valid plate it doesnt say anything. I tryied improving my prompt again and again but the flow doesnt go further from that node. Here its my promt that I have in the node that stops (that originally is in Spanish as my workflow is in Spanish): Ask the user the following question: "Can you tell me the vehicle's license plate number?" Wait for the response, then confirm: # License Plate Confirmation Repeat the license plate and make sure the customer understands it correctly. Say the numbers, number by number and letter by letter, alphabetically, in Spanish. Example: "The license plate is: one, five, three, two, B for Barcelona, ​​M for Madrid, G for Granada." If it's a letter, say the corresponding word from the following table: A of Almería B of Barcelona C of Cádiz D of Dinamarca E of España F of Francia G of Granada H of Huelva I of Ibiza J of Jaén K of Kilo L of León M of Madrid N of Navarra O of Oviedo P of Portugal Q of Queso R of Rioja S of Sevilla T of Tarragona U of Uruguay UVE. of Valencia UVEDOBLE. of Whisky X of Xenón Y of Yolanda Z of Zaragoza. #If the license plate that has been confirmed is incorrect If the user says that the license plate you have confirmed is incorrect, ask for the license plate again and confirm it again in the same way I explained before in "# License Plate Confirmation." If after reconfirming the license plate, the client says it is incorrect, repeat the process I explained in "#If the license plate that has been confirmed is incorrect." (I continue the prompt in the next message I will send due to the word cap)

I had an error with my credit card (A) and updated in the billing page to a different one (B). After I solved the problems with credit card A, I switched back to that in the billing page, and bought a few credits to make sure it was working. But the on July first, both cards, A and B, were charged with the agency plan for $500. I waited to see if the value was refunded in credit card B but it was not.

I see that campaigns can be created via the API, and within the customers section, enter the phone number and the variable values. Do I do this through the API? Is there a limit to the number of customers available through the API? Is it possible to make the call with 2,000 or more customers, or would it have to be done manually with a CSV?