and ask the ai to read out the response

but with the new tools you can do a lot of other things as well

Guys I am seeing huge difference between latency on the VAPI dashboard and when I use it on my web client. Initializing takes 5-6 seconds which is too much. Any tips on how can I reduce that? PS : Using the smallest models for fastest results, using deepgram for STT.

Mine too, 6+ seconds

Exactly See so much delays before starting conversations .

its been really absurd time for a while using vapi Ai unlike others for my use case. am using it to close free test drives for realtors butt, even after adding real sound voice plus, scripted my prompt in most human like way. still client is some-how recognizing 'maybe some Ai is talking..?' anybody have any solution to this case?

Even the most realistic voice out there will be recognized by most humans especially the more the AI talks

that seems true, actually. currently testing out vapi voices-those seems to work best for me, hopefully🤞

Interesting, I didn’t find them that good. I haven’t found anything better than elevenlabs voices yet

hmm, i don't know UR use case extactly. are you using it for outbound or else inbound? some vapi voice are not good & some are really good(like spencer)

lets chat

How do i add more Parameters to the google_calendar_tool ?

hey guys, i see that there is a lot of issue when it comes to transcription. which ones do you use in production that works best on phone calls ?

it's not a huge issue. none that i've seen are perfect, but the main issue is just with names of people and products, and spelling out letters. so you just need too make sure to handle that in some way, like asking the person to spell their name for example.

Hey guys, I built a transient assitant using vapi and deployed it on make for outbound and ibound calls. But currently when i call the assistant for inbound or it calls for outbound.. Immediately it says the first message. He stops talking. He does not read nor respond to my voice repose. What could be wrong. Note: I worked perfectly well just yesterday

Hey folks - I'm attempting to build a customer support agent. There are 3 primary parts to the call: 1. Identify the customer's issues 2. Identify which file / reference guide from the knowledge base is most relevant (pick only one) 3. Read the troubleshooting steps one-by-one I have #1-2 working, but when it gets to #3, the voice agent isn't following the troubleshooting steps from the specific reference guide in its knowledge base. Any guidance? Thank you in advance🙇🏻

Also for the "Workflows" flowchart feature, can you connect Knowledge Base files to that?

message has been deleted

Hey, create a support ticket and ping me over there

hey, can anyone tell me does vapi store the prompt if from the prompt a call is made? i lost my prompt changes, does anyone know how can i get them back

Hello All, This is Vijaya, I am new to Vapi, I am creating VAPI voice agent using APIs, how would i add system prompt while creating the agent via API, please help

Hey Vijaya You actually don’t need to add the prompt and other info to your Request body and call it very time you need it. You can create static assistant in vapi and add dynamic variables in the prompt if you need it (eg: {{first_name}} ) and mention only the dynamic variables and the assistant id in the request

You can go to call Logs to find the request that Vapi made to trigger the call but the Prompt will be in JSON Format so you may need to do some formatting to get it to its original form

Hey there Maybe try mentioning to follow the EXACT steps provided in the Knowledge base and have a section in ur prompt which provide guidance on how to use the knowledge base. Try to mention it several times and sometimes with ALL CAPS If this don’t work, let me know

am automating the Voice Agent craetion in Make.com, I am using the HTTP module to create it automatically, I need to pass the system prompt and first message as params, is there a way to pass these? or can I clone an existing agent and then pass the firstmessage and systemprompt as parameters?

can i share my script with you guys and you tell me where i need to make improvements?

Same!! Did you find a fix for this?

Yeah that’s one way to do it! Remember you can use assistantOverrides too when making calls

thanks, However I an unable to send anyting into System Prompt from the content field. Is this beacuase I am usign Free VAPI API Keys?

We've archived this channel - please use #1359790249298886667 to post your prompts! You can use threads in the forums to chat about each prompt. And if you are looking for support, please use the support forum.