Part 1: How to Build an AI Voice Agent using OpenAI Realtime API

Great work man.

Hey Bart great video just subscribed, looking forward to more great content.

good video!

superb AI Voice Agent tutorial Bart. Is it possible to cross check a booking to a sync Google calendar

Hey buddy, this is awesome! Thanks so much for sharing it with us! Can you adjust the Replit config for Vonage? Thanks again, man! 🤙🏻

Legend! Cause realtime api is hella expensive, can try deepgram agent api or livekitt by livekit which is using deepgram + 11labs but is really realtime

Great content Bart! One tip for you, I'd use your environment file for your project to store sensitive constant variables like your Make Webhook URL so people don't try copying your code and spam your webhook endpoint. And/or implement some webhook public + private secret system to authenticate webhook requests. That way, you don't get random people on the internet trying to access your APIs and systems :) Keep up the good work.

I've heard this is very expensive to run.

GOAT 🐐
Keep it up

OpenAI is releasing swarm soon. Hopefully it helps take off all this load to be able to easily and more fluently configure such things.

Still way too complicated. I’ll wait.

top, muito bom, funcionou a v1 aqui, vou testar essa v2 no n8n

Great video! Nice pace and great way to explaining the technical stuff

Hi Bart, yes please could you make a video of how you built the automation backend, I’m very keen to learn more from you.

Many Thanks :) could you please share your Make automation ?

Thanks Bart!
I'm looking forward to more great content.

HELPFUL! Thanks brotha

Amazing content Bart, thank you

Great video! Would you be open to sharing the Blueprint for your Make configuration?

Thanks, amazing an it really helps a lot, but I don't know y after I run it it start ti reply to it self without my input :((

Excellent material! Congratulations. Keep going 💪

Great work! Insanely useful!

this is great video! Thank you!

Hello Bart! Thank you very much for the guide, it's great! I have a question about transcription. I want to set the language to Polish because it often doesn't detect words. I tried to set leanguage pl under whisper-1, but then I have disruptions during the conversation. Thank you for your reply!

Thank you for this, you explained the code extremely well, even for me as a beginner of NodeJS! I have a question: I do not have replit, but I usually use Google Cloud for things like this. Is it possible to deploy this NodeJS app using Google Cloud somehow? I read about The Google App Engine, however as far as I read the Standard environment (which is free up to 28h a day) does not support WebSockets. Does this mean that this would not work in this case?

Great Video Bart! Keep them up.

Great stuff! Insanely useful!

did you say hello legends?? the only legend here is you bruh lol

Bart, where are you connecting to any data sources or any context for the RAG? I don't see it in the index.js. Looking to extend this and see if I can set this up to a knowledgebase to properly answers questions with factual information from the knowledgebase. Thank you, this is amazing.

Amazing work by you I’ve learned alot

Great tutorial!

This was just the video I was looking for. Many thanks.

This is awesome. I already made v1, but v2 i followed to too the script, it must be cause I'm on a trial that it won't make the call to webhook / media stream cause everything else work. Any suggestions?

Possible to post equivalent code in Python?

Hey. Does bland and vapi use realtime or they use something like your other video of deepgram as the speech analysis / transcript and then chat gpt api calls for answers ?

Hi Bart, this is super awesome really helped a lot. Quick question i keep on getting error on twilio invalid schema 12200 so basically when i call nothing happens. It happens randomly not sure what is causing it. Is it the firstmessage ?

Great video. Do you plan to create one that integrates with openAI assistant directly (or may be pinecone) instead of through make?

Hi, Bart! Great tutorial! You are one of few on YT who explain how to deploy it with Twilio. So, that's how I find you ;) Could you be so kind to export .json of Make scenario, so we can test the entire automation. It'll be very helpful. You can add it in your resources under this video. Otherwise, we have incomplete solution. Thanks again. I hit subscribe, so waiting for new updates 😉!

Great Job dude!

Amazing content to get any one started. Thanks!

Excellent Bro

😂 it's funny you edited the video to hide the latency issue you are facing. Be transparent brother. Otherwise you are on the scammer road.

DO you have one using python?

The realtime api is still buggy. It is diffcult to recite numbers such as your phone number if you are building a medical appointment app

Thanks for video , maybe need all those
1) RAG
2) Function calling
3) Caller history
4) Call forwarding

Great project!

is there
any way to use the configiration from
make?

Awesome. Thanks man

You have made the explanation so technical. Very hard to understand 😢😢😢