@@jannismoore actually I ran into an issue after completing the setup - I get an error "Account doesn't exist: Connection not found 'google-sheets:2132426'." I tried to fix it following a few suggestions, like resaving the modules etc. Still can't get it to work... any suggestions you can send my way?
It’s probably a better alternative than using the mailbox feature as of now. That’s easily possible to build within the same flow setup. Interested in seeing this being integrated?
Yes very interested, also will be nice if from a website formulary or other similar source feed the google form that trigger make by simple adding a new row. Thanks for sharing your knowledge is impressive.
I cannot get the endcall function to work within a Make HTTP module. Do you have a video showing this working? Eg the assistant ending the call rather than the person… thanks
@@jeremya4160 It also runs through the end-of-call-report message type. It has a field set called endedReason or something like that which determines the reason for the call ended. You can use that one
Hi @Jannis Moore, Thanks for this video. It's really great to always come back to your resources. Please I need your help. I ran this outbound cold caller script but somehow I'm not able to have it execute. Could I book a session with you? Thanks
If the issue ia about Vapi not calling: Just check if you have an outbound number as well. If you don't have an outbound number attached to vapi then the AI can't call you. If you only have an inbound number then you can only call the AI through that number
Plenty. My next video is a complete guide on creating an inbound setup using transient-based Assistants. It's going to be a huge tutorial so in case you look for something specific, feel free to share it here
@@jannismoore Sounds good - looking forward to that as well. For the inbound calls, I was thinking about the plumber that came to my house one day to do some work - he was there probably 30 minutes and took probably 5 calls. English was not his first language even though I had no issues understanding him, but on the phone, he had to repeat a few things, and the questions being asked were pretty much the same based upon his response - like when can you be here and how much is it going to cost etc. To help someone like this would be good. I think when he is at a job, he would forward his number to the virtual number, so when someone calls, they get the assistant who would then track everything (google sheet) and at the end of the call, a text is sent to the real number, so that when he gets in the truck, he has all the details. He can then call immediately and also has the google sheet to refer to at a later date. How does that sound
Hi Jannis - everything worked until the very last module. google sheet row update. Am getting the following error - Missing value of required parameter 'rowNumber'.
@@iainhmunro Definitely interesting. In the end you should think about if that potential prospect would actually see the value and that you can get enough money to make it worthwhile for both. This can be tricky in some industries as the costs are still pretty high and some people prefer to do calls themselves instead of having the simplicity of calling an AI agent as the effort doesn't bother as much as the lack of money
@@iainhmunro Hmm, if you run "Play" once, you will see the actual data flow in the little bubbles at the top right of each module. There you can actually check out the Update Row item as it will show you the actual response. It might be that the number you tried to update didn't exist at the time of execution (or it was maniupulated)
You can leverage the second make.com scenario that I shared (Dynamic Cold Caller Sync Data). If you bring your own Twilio phone number, you can add a Twilio action within the scenario at the end and then send an SMS that way. If you leverage other bulk SMS services, I suggest whether or not they have an integration with make.com or not. If you need more details, feel free to let me know!
My buddy Brendan released an awesome video on this - feel free to check it out here: for those that haven’t seen: ruclips.net/video/sGF_-7pDcXA/видео.html
Totally. You can simply build them out visually and they will directly integrate with this automation template. You then only need to build the actual scenario for the function call to do whatever the custom function is supposed to do.
Hey Jannis, thanks for the video tutorial and resources. I followed your instructions to the tee but when I run the Dynamic cold caller I dont get a call. I put my number in to test it but didn't receive a call. I updated the modules you said in the video like assistant id, phone id, google sheet. Any idea what could be hindering the call? I did update the sync data scenario and updated the assistant with the URL of the webhook.
Awesome tutorial Jannis!! Thanks!! Btw, how would you go about adding sms and/or email integration?? So after the call, the lead gets an email or a text.
Thanks for sharing, Jannis. I would like to know if there is any reference to the tutorial for setting up airtable? I have been using airtable in conjunction with make.
@@jannismoore Glad to hear from you so quickly, Jannis. I am looking forward to whether it can become the next video. My two small suggestions are to combine transcription or summary with ChatGPT-4o to generate suggestions for the next step of following up with customers, which is a good guide for sales staff. In addition, if the overall customer impression of cold calls can be scored by AI, it will also help to improve conversion rates. I hope my ideas can give you some inspiration.
hi, thanks for the tutorial, at your companys website's login page, the register button is not visible due to white color. It would be better if you change it to something colorful so people can register easily
You can do it without as well, but then you will only have a transcript and no other data that may be relevant from the transcript for further programmatic use. If you want to access some specific data from a transcript, you would have to look through each transcript manually, but using the OpenAI classifier I included, you can extract that data and check against it programatically. You won't need that if you look at only 1-2 calls a month, but if you do more than that, this comes quite handy.
@@jannismoore Thanks. One last question: regarding the 'dynamic cold caller Sync Data', in the third function OpenAI, where the transcript is retrieved with 6.message.transcript, it does not seem to be active on my setup. In the video, it shows a red background object, but on mine, it shows a white/transparent background. The webhook was set up as instructed. Do you have any idea why this might be happening? Thanks again.
@@dotfrontllc2300 White means that the field wasn't available in the request you sent. This may be because your Vapi assistant triggers in different events when testing it. You probably should remove all of the other events within the serverMessages section except of the end-of-call-report
@@jannismoore Thank you for your prompt reply. I removed all as instructed, including ClientMessages, and selected their defaults again. I made sure that the data structure was successfully determined. However, nothing has changed; it is still not available. recommendations? thanks again...
I get a statusCode 400 when I run it. Second issue is that I have registered at JSONaut twice with two different email addresses and not gotten a confirmation email for either so I cant log in and get the api key. thanks in advance for any help you can provide.
Great tutorial! So, to sell a complete and professional solution, could we use GHL instead of Google sheets? I feel like that can make the whole progress more seamless and automate things like e-mail and whatsapp follow up. Do you have a tutorial on that?
Hey a small Doubt , I haven't got a VAPI premium and I'm just using a VAPI Trial Number, I can see the make scenario getting executed successfully but I still get no call from the AI What's likely the issue here? UPDATE: IN the output bundle of the https node (last) I see some 400 error status
Hey, I don't even think Vapi has a premium acc - it's kinda all the same. If you see a 400 error, you most likely also see some more details in the request. Can you share those?
Great video, well done, for some reasons it's not working for me, I did follow all your video instructions, for the second scenario (Dynamic cold caller sync data I couldn't save, it ask me to enter a Webhook. When I test, no calls are coming in, I entered my own number in the Google sheet, I'm using my Twilio account, any ideas? Thanks
Can you please tell me how to add new questions and it should display on google sheet like you added two columns want to sell and more details about property
Check the bubble on the top right of your HTTP module - it most likely shows a 400 error including more detials on what's wrong. You should also watch that video to understand how you can actually debug scenarios: ruclips.net/video/zdWS9sAd6PQ/видео.html
@@jannismoore would you be able to dive more into dyamic tags a calling someone by their 1st name on the phone can sound unprofessional especially older generation- would the bot recognise "Mrs, Smith", i did get a bit lost with the hooking everything up with make had to pause & rewind lol- good video though
@@mattbaker8267 Hey Matt! You can add any kind of dynamic tags to it. I will soon release a complete tutorial on how to create an outbound phone calling agent, which will also cover the dynamic tags more in-depth. As for now: You can define the additional data on how you want to address the user within your lead list and then create a new tag within the dynamic variables step inside of the make.com scenario. Then you can use that tag within your transcipt inside of make.com
@@jannismoore Hi Again Jannis- I seem to be having issues with the 2nd automation its not liking the webhook Error Failed to connect to remote server. Failed to execute 'clone' on 'Response': Response body is already used
@@mattbaker8267 That's an interesting one I've never seen. Upon Googling it, it seems this comes when connecting a third party app? I suggest re-saving tour scenario to a new one (you can copy modules). There might be also a browser integration interfering with this.
@jannismoore - This is AWESOME!!! However I've tried configuring a number of times but keep getting an error on the last google sheets in the Cold Caller Sync Data: Error Validation failed for 1 parameter(s). - Missing value of required parameter 'rowNumber'. What am I missing here? I'm sure its really simple, I just cant see it!
@@jannismoore I've used it as is after downloading it with columns A to E for Google sheet Cold Calling Database. I've also tried modifying it but either way cant get it to work.
actually I went back to ground zero, deleted all workflows, connections, webhooks and sheets and started all over. I think there was an issue with the worksheet connection refreshing correctly. If I did make changes to the worksheet which steps in the workflow would I need to update?
@@jjwprotozoa This depends what exactly you change. If you change the sheet name, you need to select the new sheet name from the dropdown. If you change columns, you may have to map them again to the right order. It's best to add anything new at the end of the last filled column to not cause issues with the previous ones.
Hey Jannis, great video as usual but I have an issue I did the same process but not getting a call and at last step which Is a HTTP post request I am getting 400 error.
Hey! If you're getting a 400 error, you should definitely check out the data within the response. Vapi send back an error message in the HTTP block which is where you can see more details about what is wrong
Hey! You probably need to create a new webhook URL in the trigger of the scenario and then add that one to the Server URL within Vapi. This is account-specific, that's why make.com doesn't import it
@@jannismoore Hey thanks for the reply! I made the new url, but cant seem to find where to add it to the server url in Vapi... Sort of hard to troubleshoot when I didnt really go over how you set it up haha
Hey Jannis, thanks for the tutorials. I had already a scenario and a setup before I found your video. Yours is easier ! ah. But, when I try to lunch a call out, it gets disconnect right away and it hungs up. I'm using an import from Twilio. Any ideas ?
@@manouky One more thing you can try is to check if your Twilio Number is funded and not in trial state. This can affect the functionality (Given you have imported a Twilio number)
@@jannismoore Hey ! So it turns out that after researching, 11labs needs you to have a paid subscription on top of what you pay at Vapi. How dumb. It seems to work for others, no pro member needed. However, for OpenAI API model, i got a pipeline error, i have to use Antropic to work. PS : I have premium accounts on both, but not sure why I have a pipeline error on OpenAI.. cheers
Hi! In the google sheets last scenario, show an error: The operation failed with an error. 403: PERMISSION_DENIED - The caller does not have permission. - what should i do?thanks
403: PERMISSION_DENIED - The caller does not have permission Origin Google Sheets Automatic error handler If you want to handle this error automatically, choose one of the following options. This will create a new error-handler route in your scenario. You can then expand the route in any way you like.
amazing stuff man!] I only have one problem with the phone number. I live in Spain, so I can´t buy a PN directly. So, if I want to export one from Vonage, it will charge me a monthly fee for using it, plus what Vapi is gonna charge me for the use of the Assistant. Is there a way to pay for a PN only once..and then pay for the assistant as you use it. Thank a lot!
Hey! You can purchase a number through Vapi. It's still not a one time fee, but only $2 recurring + the usage fee whenever you have a call. Mostly any provider charged you a monthly fee on their numbers.
@@jannismoore Thx for the response! The thing is Vapi is not offering Spanish numbers at the moment (only US and Canada). So, I would need to import them from Twilio or Vonage. I´m figuring that out right now..
I am not receiving an email conformation for my Jsaonaut account and so I can't sign in, what should I do. I also tried reseting my password but still the same issue persists, I receive no email.
Hi Jannis, I tried this and followed directions closely.. It will not call the phone number. Not sure what I'm doing wrong. I tried it with a 1 and without a 1.
If you purchase the phone number via Vapi, I guess it should support the number out of the box. If you import your own Twilio number, you need to whitelist your country code in the Geo permissions of Twilio
You need to run the scenario once manually and see the responses - those should tell you all the details about potential errors. Once you check them, you'll most likely get a better picture of what's wrong. Feel free to let me know about the details of the error then
Hi, man are you German? I am from Norway. Anyway I was hoping you could point me in the directions of a very good norwegian LLM, I am using Norbert at this moment, and I have not yet found a good TSS who can turn text into speach in a natural sounding voice. I am using vapi azure nb-NO list and well they dont speak very good norwegian, and it seems vapi do not handel it so well either. That is why I need also a great prompt to make my Agent to sound natural. Could you help?
Hey, I barely worked with a Norwegian language, but I guess azure has the biggest variety. You might want to look through their Voice Gallery to find one that seems suitable. Even though it's not that fluent yet, I'm sure you'll see increased accuracy with time. Vapi is still the best solution as you can later on switch once there's something better on the market. The prompt per se doesn't help with the pronounciation. It might make a difference with the way you structure sentences and punctuation, but generally this affects it only in minor ways.
429 usually indicates that too many requests have been sent. This looks like a rate limit error. Not sure about their internal error handling, but it could be a problem. Does that error appear as well when you only run the OpenAI module within the make.com scenario?
@@vishwasdhabhai229 It may be that it is a demo account. I suggest to look into your usage to see if that is increased. Otherwise you can connect it just like I've shown it in the video. It's pretty straight forward so you can simply follow the guide step by step
@@vishwasdhabhai229 Not necessarily. I haven't had that issue so far. It's more of a guess that is probably faster than waiting for an answer of the OpenAI support.
I'm not a legal adviser, but I assume that if you have gathered their permission in some way (marketing sign ups within your T&C), it should be ok. Calling someone completely out of context is certainly not a thing you want to do
@@EdemirLandau You have to be careful which company information you use. Anything in the Impressum isn't allowed, as that isn't made public voluntarily, but as a legal requirement.
@@iainhmunro Of course lol. I put the phone Id into the work flow. Someone keeps telling me I have to hard code +1 in the workflow.. Not sure what that means.. Ill keep trying
@@iainhmunro I finally got it. I had to put in the exact Google sheet id as it was pulling from a different one than I selected. Thanks for reaching out
The latency is realistically around 800 to 1000ms. I’m about to release more long form content about inbound AI Agents with Vapi and I’ll make sure my editor won’t cut any potential pauses
@@SlaiyAi It definitely is. We've build not only AI Agents on top of it, but whole SaaS tools. Vapi is my go to solution and you can adjust things depending on what's important to you. You can lower the delay more effectively than with platforms like bland.ai. If you have specific questions, feel free to ask them at any time.
Hi Jannis. Love this video thank you for sharing - 400: INVALID_ARGUMENT - Unable to parse range: 'Sheet1'!undefined2:undefined1000000 Can you advise what the issue here is?
It seems your row is undefined. Maybe the Google Sheet you copied has been altered. I suggest going into the make.com scenario Google Sheet integration and see what comes through when you run it once.
@@jannismoore Thank you for getting back to me just earned a subscriber! if I wanted to change the copy on the Vapi and one section in the script instead of sell do you want to be a partner would that have an effect? If so where would I go to update that?
Hey, just so you know Cold calling using Ai is Illegal in the US. The business doing the call needs to have explicit consent to call the individual using a system like this.
Having permission makes it not illegal, just like you’re saying it yourself. I think it’s obvious that you cannot just call random numbers without having any sort of permission. If you’ve a proper privacy policy and T&C in place, I don’t see why this should be a problem. Simply allow users to unsubscribe from cold calling, just like you’d do it with emails. I’m obviously not a legal advisor, but this is how I personally address this. There’s probably as well a differentiation between corporate and personal numbers.
I can make the best tutorial, provide predefined templates AND show it on the screen, and people still manage to mess it up. It’s just the nature of things, especially if someone tries to do things that they haven’t done before. It seems you’re in the same position, so going through the comments and sharing nothing constructive is a clear representation of the position you’re in. I read comments, so with that you just lost your option of actually getting a helpful answer, but instead a lecture to something that isn’t even partially important
@@jannismoore that is fair of you to say, so instead i decided i will actually give it a shot and not be deconstructive to you and your channel. Now having said that i believe I am at a roadblock where at the module does not work. I believe it may be to do with the the part of your video at 8:21 and it says to enter a URL into the webhook. What URL is this?
@@ammanrawjirealestate well done. It talks about the webhook URL of the initial module you see within the Make.com scenario. When you click on the red webhook module, you can generate a URL (You need to do this anyways before you can save the scenario). Now this URL can be added within your account as an account-wide URL which is then used to receive any of your server messages and assistant requests (for transient assistants). Now initially this was the only way of doing it, but now you can also assign a webhook URL (also called serverUrl) to a phone number directly. Here's the direct link for that (you'd need to update this via an API request: docs.vapi.ai/api-reference/phone-numbers/update#request.body.serverUrl) If you're okay having it account wide, you can continue the tutorial as you see in the video while adding the webhook URL from the webhook module of the scenario into the Server URL field located within your Vapi dashboard under Settings > Server URL
I’ve informed my editor already. If you consider it slow right from the start, I assume it’s more of a mindset issue than the actual technology. Speed always depends on the right prompting, LLM, and other transcriber services. Vapi is benchmarking it’s LLMs directly during a call so it can choose the fallback model that is fastest, giving you an advantage over other providers. At this point in time, building voice agents is always connected with a trade-off which requires you to balance speed and quality, but from what I’ve seen so far, you can do great with the right prompting and setup. Each technology provider is evolving rapidly and bringing in new features that make the conversation more human-like. Look for example at back-channeling or static function messages.
It's actually not slow at all. If you use models such as Groq for instance and the correct transcriber etc the responses are pretty much instantaneous. Also bare in mind if you have ever spoke to anyone from a call centre or a cold call, every human will pause between answers. Usually as they are writing something on their system as they go, or running through a script so there is always a pause. These services are actually much faster I would say.
Hi @jennis moore …..it was an amazing video but I still have a doubt on how to buy a number or get it from twilio. Please explain the process of numbrr a little bit. If anyone of you know thud please help me
Hello, Can you please tell me how to add new questions and it should display on google sheet like you added two columns want to sell and more details about property
You can do that by adding them via the Structured data inside of Vapis analytics tab. You can then add those fields within make.com to the required columns
Amazing - thanks for sharing
That was awesome and exactly what I was looking for!! Thank you!!!
Glad it was helpful! Is there anything else you’re struggling with?
@@jannismoore actually I ran into an issue after completing the setup - I get an error "Account doesn't exist: Connection not found 'google-sheets:2132426'." I tried to fix it following a few suggestions, like resaving the modules etc. Still can't get it to work... any suggestions you can send my way?
@@reubencano5171 I suggest verifying the Google Sheet connection and refreshing the columns/fields
Whoa, amazing man.
Best video so far - after all this process if no answer - you can set up to try multiple times until success
It’s probably a better alternative than using the mailbox feature as of now. That’s easily possible to build within the same flow setup. Interested in seeing this being integrated?
Yes very interested, also will be nice if from a website formulary or other similar source feed the google form that trigger make by simple adding a new row. Thanks for sharing your knowledge is impressive.
@@EdemirLandau Just added that to my list. Thanks for the input!
This is great, thank you! I think a video about using custom functions would be great!
It's already planned :) Anything specific you want me to cover for function calling?
I cannot get the endcall function to work within a Make HTTP module. Do you have a video showing this working? Eg the assistant ending the call rather than the person… thanks
@@jeremya4160 It also runs through the end-of-call-report message type. It has a field set called endedReason or something like that which determines the reason for the call ended. You can use that one
Hi @Jannis Moore, Thanks for this video. It's really great to always come back to your resources. Please I need your help. I ran this outbound cold caller script but somehow I'm not able to have it execute. Could I book a session with you? Thanks
WOW.. GOOD LOOKING OUT!
If the issue ia about Vapi not calling: Just check if you have an outbound number as well. If you don't have an outbound number attached to vapi then the AI can't call you. If you only have an inbound number then you can only call the AI through that number
Hi...I m ur new subscriber from pakistan
Love it !
Hey I get hook not found device not found when I turn it on?
Thanks Jannis - that helps out a lot. I will give this a whirl this week. Have you got anything built yet for doing an inbound scenario ?
Plenty. My next video is a complete guide on creating an inbound setup using transient-based Assistants. It's going to be a huge tutorial so in case you look for something specific, feel free to share it here
@@jannismoore Sounds good - looking forward to that as well. For the inbound calls, I was thinking about the plumber that came to my house one day to do some work - he was there probably 30 minutes and took probably 5 calls. English was not his first language even though I had no issues understanding him, but on the phone, he had to repeat a few things, and the questions being asked were pretty much the same based upon his response - like when can you be here and how much is it going to cost etc. To help someone like this would be good. I think when he is at a job, he would forward his number to the virtual number, so when someone calls, they get the assistant who would then track everything (google sheet) and at the end of the call, a text is sent to the real number, so that when he gets in the truck, he has all the details. He can then call immediately and also has the google sheet to refer to at a later date. How does that sound
Hi Jannis - everything worked until the very last module. google sheet row update. Am getting the following error - Missing value of required parameter 'rowNumber'.
@@iainhmunro Definitely interesting. In the end you should think about if that potential prospect would actually see the value and that you can get enough money to make it worthwhile for both. This can be tricky in some industries as the costs are still pretty high and some people prefer to do calls themselves instead of having the simplicity of calling an AI agent as the effort doesn't bother as much as the lack of money
@@iainhmunro Hmm, if you run "Play" once, you will see the actual data flow in the little bubbles at the top right of each module. There you can actually check out the Update Row item as it will show you the actual response.
It might be that the number you tried to update didn't exist at the time of execution (or it was maniupulated)
Great tutorial 👍 question: how would I then send a text message to the prospect at the end of the call?
You can leverage the second make.com scenario that I shared (Dynamic Cold Caller Sync Data).
If you bring your own Twilio phone number, you can add a Twilio action within the scenario at the end and then send an SMS that way. If you leverage other bulk SMS services, I suggest whether or not they have an integration with make.com or not. If you need more details, feel free to let me know!
@@jannismoore I really appreciate the detailed instructions and the time you took to reply. Thank you!
Can you do a video sharing how to link this work flow to GHL booking calendar. That would be GOLD!!!
My buddy Brendan released an awesome video on this - feel free to check it out here: for those that haven’t seen: ruclips.net/video/sGF_-7pDcXA/видео.html
Hey Jannis, this was awesome. I want to know one thing can we add custom functions in this like appointment booking function?.
Totally. You can simply build them out visually and they will directly integrate with this automation template. You then only need to build the actual scenario for the function call to do whatever the custom function is supposed to do.
@@jannismoore Thank you so much
@@jannismoore Can you let me know where we can add the webhook url i am unable to find server url in vapi.
Hey Jannis, thanks for the video tutorial and resources. I followed your instructions to the tee but when I run the Dynamic cold caller I dont get a call. I put my number in to test it but didn't receive a call. I updated the modules you said in the video like assistant id, phone id, google sheet. Any idea what could be hindering the call? I did update the sync data scenario and updated the assistant with the URL of the webhook.
Awesome tutorial Jannis!! Thanks!! Btw, how would you go about adding sms and/or email integration?? So after the call, the lead gets an email or a text.
It depends which country you're talking about. If it's simply for the US, you can do that with Twilio/ + Mailgun/Sendgrid
Thanks for sharing, Jannis. I would like to know if there is any reference to the tutorial for setting up airtable? I have been using airtable in conjunction with make.
I don't have a tutorial for that available as of now, but if you want, I can get that added to my list.
@@jannismoore Glad to hear from you so quickly, Jannis. I am looking forward to whether it can become the next video.
My two small suggestions are to combine transcription or summary with ChatGPT-4o to generate suggestions for the next step of following up with customers, which is a good guide for sales staff. In addition, if the overall customer impression of cold calls can be scored by AI, it will also help to improve conversion rates.
I hope my ideas can give you some inspiration.
hi, thanks for the tutorial, at your companys website's login page, the register button is not visible due to white color. It would be better if you change it to something colorful so people can register easily
It doesnt call when I have more than 1 number row? Im usung Zoho Creator as the spreadsheet
Hello instead of using Google sheets can this exact thing be done with GHL? Could you make a tutorial on that
jannis please google sheet is not showing anything
Great video! Question: Why do we need OpenAI if Vapi uses a transcriber?
You can do it without as well, but then you will only have a transcript and no other data that may be relevant from the transcript for further programmatic use.
If you want to access some specific data from a transcript, you would have to look through each transcript manually, but using the OpenAI classifier I included, you can extract that data and check against it programatically.
You won't need that if you look at only 1-2 calls a month, but if you do more than that, this comes quite handy.
@@jannismoore Thanks. One last question: regarding the 'dynamic cold caller Sync Data', in the third function OpenAI, where the transcript is retrieved with 6.message.transcript, it does not seem to be active on my setup. In the video, it shows a red background object, but on mine, it shows a white/transparent background. The webhook was set up as instructed. Do you have any idea why this might be happening? Thanks again.
@@dotfrontllc2300 White means that the field wasn't available in the request you sent. This may be because your Vapi assistant triggers in different events when testing it. You probably should remove all of the other events within the serverMessages section except of the end-of-call-report
@@jannismoore Thank you for your prompt reply. I removed all as instructed, including ClientMessages, and selected their defaults again. I made sure that the data structure was successfully determined. However, nothing has changed; it is still not available. recommendations? thanks again...
why can't we connect it with Readymode dailer is that possible??
If it supports SIP, you can connect it to most of the major service providers, including Vapi
Is it possible to train the cold caller and give it a script of responses to give when the person says certain things?
Yep, you can add that to your prompt instructions
I get a statusCode 400 when I run it. Second issue is that I have registered at JSONaut twice with two different email addresses and not gotten a confirmation email for either so I cant log in and get the api key. thanks in advance for any help you can provide.
Cool! Thank you for sharing this. Which screen recording are using? Wondering how it follows the mouse? Or was that postproduction?
Hey! It's called Screen Studio
@@jannismoore Thanks man
the number that the vapi ai will call how does that work can you tell me again please sir
Great tutorial! So, to sell a complete and professional solution, could we use GHL instead of Google sheets? I feel like that can make the whole progress more seamless and automate things like e-mail and whatsapp follow up. Do you have a tutorial on that?
I'll build the GHL version if you want to collab on it.
Hey a small Doubt , I haven't got a VAPI premium and I'm just using a VAPI Trial Number, I can see the make scenario getting executed successfully but I still get no call from the AI
What's likely the issue here?
UPDATE: IN the output bundle of the https node (last) I see some 400 error status
Hey, I don't even think Vapi has a premium acc - it's kinda all the same.
If you see a 400 error, you most likely also see some more details in the request. Can you share those?
Do you use other no code apps like Clear Talk?
I do use other apps, but I've never heard of this one.
Great video, well done, for some reasons it's not working for me, I did follow all your video instructions, for the second scenario (Dynamic cold caller sync data I couldn't save, it ask me to enter a Webhook. When I test, no calls are coming in, I entered my own number in the Google sheet, I'm using my Twilio account, any ideas? Thanks
I have the same problem, not receiving any calls... Did you figure out what was the problem ?
Can you please tell me how to add new questions and it should display on google sheet like you added two columns want to sell and more details about property
I have quite a couple of videos on my channel - for sure I covered that somewhere
I'd love to get this working but I never got a call when i set this up. No error messages when running the first scenario just never calls.
Check the bubble on the top right of your HTTP module - it most likely shows a 400 error including more detials on what's wrong.
You should also watch that video to understand how you can actually debug scenarios: ruclips.net/video/zdWS9sAd6PQ/видео.html
@@jannismoore Thanks! I'll give it a try!
I never received a call also. Followed all the steps and no error. Would it have anything to do with the Openai API running out of tokens? @jannis
greate video
Glad you enjoyed it!
The hub knowledge links dont work
What’s the error you face? We can resolve the URLs at our end
Would you be able to do a full tutorial, as this was a bit quick- mainly connecting make with everything😇
I would and I will. It's already on my video timeline :)
Anything specific you want to see being covered?
@@jannismoore would you be able to dive more into dyamic tags a calling someone by their 1st name on the phone can sound unprofessional especially older generation- would the bot recognise "Mrs, Smith", i did get a bit lost with the hooking everything up with make had to pause & rewind lol- good video though
@@mattbaker8267 Hey Matt! You can add any kind of dynamic tags to it. I will soon release a complete tutorial on how to create an outbound phone calling agent, which will also cover the dynamic tags more in-depth.
As for now: You can define the additional data on how you want to address the user within your lead list and then create a new tag within the dynamic variables step inside of the make.com scenario. Then you can use that tag within your transcipt inside of make.com
@@jannismoore Hi Again Jannis- I seem to be having issues with the 2nd automation its not liking the webhook
Error
Failed to connect to remote server. Failed to execute 'clone' on 'Response': Response body is already used
@@mattbaker8267 That's an interesting one I've never seen. Upon Googling it, it seems this comes when connecting a third party app? I suggest re-saving tour scenario to a new one (you can copy modules). There might be also a browser integration interfering with this.
What if I want to use Anthropic can I switch the data from Chatgpt into the claude
Totally! This assistant setup is compatible with all of Vapis features (As of the time when I made this video)
@jannismoore - This is AWESOME!!! However I've tried configuring a number of times but keep getting an error on the last google sheets in the Cold Caller Sync Data:
Error
Validation failed for 1 parameter(s).
- Missing value of required parameter 'rowNumber'.
What am I missing here? I'm sure its really simple, I just cant see it!
Hey! Have you altered the columns of sheet of the Google Sheet?
@@jannismoore I've used it as is after downloading it with columns A to E for Google sheet Cold Calling Database. I've also tried modifying it but either way cant get it to work.
actually I went back to ground zero, deleted all workflows, connections, webhooks and sheets and started all over. I think there was an issue with the worksheet connection refreshing correctly. If I did make changes to the worksheet which steps in the workflow would I need to update?
@@jjwprotozoa This depends what exactly you change. If you change the sheet name, you need to select the new sheet name from the dropdown. If you change columns, you may have to map them again to the right order. It's best to add anything new at the end of the last filled column to not cause issues with the previous ones.
@@jannismoore i have the same situation: Missing value of required parameter 'rowNumber'.
Hey Jannis, great video as usual but I have an issue I did the same process but not getting a call and at last step which Is a HTTP post request I am getting 400 error.
Hey! If you're getting a 400 error, you should definitely check out the data within the response. Vapi send back an error message in the HTTP block which is where you can see more details about what is wrong
Legend
Hey, getting an error in make that the webhooks part is empty, but I havent altered anything?
Hey! You probably need to create a new webhook URL in the trigger of the scenario and then add that one to the Server URL within Vapi. This is account-specific, that's why make.com doesn't import it
@@jannismoore Hey thanks for the reply! I made the new url, but cant seem to find where to add it to the server url in Vapi... Sort of hard to troubleshoot when I didnt really go over how you set it up haha
Hey Jannis, thanks for the tutorials. I had already a scenario and a setup before I found your video. Yours is easier ! ah.
But, when I try to lunch a call out, it gets disconnect right away and it hungs up. I'm using an import from Twilio.
Any ideas ?
Hey, what's the message you hear when you try to call it?
@@jannismoore thx for quick reply. Nothing, it rings on my cell, I pick up, then it hangs up.
Argh lol. I tried your scenario over and over, and whenever it calls my number, I answer and it hangs up right way.
@@manouky One more thing you can try is to check if your Twilio Number is funded and not in trial state. This can affect the functionality (Given you have imported a Twilio number)
@@jannismoore Hey ! So it turns out that after researching, 11labs needs you to have a paid subscription on top of what you pay at Vapi. How dumb. It seems to work for others, no pro member needed.
However, for OpenAI API model, i got a pipeline error, i have to use Antropic to work. PS : I have premium accounts on both, but not sure why I have a pipeline error on OpenAI..
cheers
so i need a number to make this work i cant use vapi directly
Hi! In the google sheets last scenario, show an error: The operation failed with an error. 403: PERMISSION_DENIED - The caller does not have permission. - what should i do?thanks
403: PERMISSION_DENIED - The caller does not have permission
Origin
Google Sheets
Automatic error handler
If you want to handle this error automatically, choose one of the following options. This will create a new error-handler route in your scenario. You can then expand the route in any way you like.
Why did you use open ai as the provider and not vapi?
Vapi is a setting for their blocks, which isn't fully functional yet
amazing stuff man!]
I only have one problem with the phone number. I live in Spain, so I can´t buy a PN directly.
So, if I want to export one from Vonage, it will charge me a monthly fee for using it, plus what Vapi is gonna charge me for the use of the Assistant.
Is there a way to pay for a PN only once..and then pay for the assistant as you use it.
Thank a lot!
Hey! You can purchase a number through Vapi. It's still not a one time fee, but only $2 recurring + the usage fee whenever you have a call.
Mostly any provider charged you a monthly fee on their numbers.
@@jannismoore Thx for the response!
The thing is Vapi is not offering Spanish numbers at the moment (only US and Canada). So, I would need to import them from Twilio or Vonage. I´m figuring that out right now..
If we have a phone number already can we use it?
It it's a Twilio or Vonage number, you definitely can
I am not receiving an email conformation for my Jsaonaut account and so I can't sign in, what should I do. I also tried reseting my password but still the same issue persists, I receive no email.
Did you check your spam folder? If you use a custom mail server - you might also want to look at your spam policies.
Hi Jannis, I tried this and followed directions closely.. It will not call the phone number. Not sure what I'm doing wrong. I tried it with a 1 and without a 1.
Jannis, nevermind. I got it to work! Don't know what happened. I went over everything again and I think I had the wrong api key for Vapi. Thanks!
Glad it worked for you!
Any idea how I can test it by calling myself, but I do live in Europe with the phone country code (+32) ?
If you purchase the phone number via Vapi, I guess it should support the number out of the box. If you import your own Twilio number, you need to whitelist your country code in the Geo permissions of Twilio
Hello, Brother it is not working i tried it the data is not showing on google sheet. Can you please help me with it?
You need to run the scenario once manually and see the responses - those should tell you all the details about potential errors. Once you check them, you'll most likely get a better picture of what's wrong.
Feel free to let me know about the details of the error then
@@jannismoore Not working. I tried everything accurately but the status is not changing and also the two questions on Google sheet are not displaying.
Hi, man are you German? I am from Norway. Anyway I was hoping you could point me in the directions of a very good norwegian LLM, I am using Norbert at this moment, and I have not yet found a good TSS who can turn text into speach in a natural sounding voice. I am using vapi azure nb-NO list and well they dont speak very good norwegian, and it seems vapi do not handel it so well either. That is why I need also a great prompt to make my Agent to sound natural. Could you help?
Hey, I barely worked with a Norwegian language, but I guess azure has the biggest variety. You might want to look through their Voice Gallery to find one that seems suitable.
Even though it's not that fluent yet, I'm sure you'll see increased accuracy with time. Vapi is still the best solution as you can later on switch once there's something better on the market.
The prompt per se doesn't help with the pronounciation. It might make a difference with the way you structure sentences and punctuation, but generally this affects it only in minor ways.
Can you help me for fix my caller I can’t making the cold calls , I could pay you
Feel free to reach out via our website integraticus.com
I have 5 dollars as a free given openAI Credits. and it's not used but the error 429 error is been shown. is this normal ?
429 usually indicates that too many requests have been sent. This looks like a rate limit error. Not sure about their internal error handling, but it could be a problem. Does that error appear as well when you only run the OpenAI module within the make.com scenario?
@@jannismoore yes it's when i run the sync data scenario. what's the fix? Also how to connect both the scenarios? sync data and the other one?
@@vishwasdhabhai229 It may be that it is a demo account. I suggest to look into your usage to see if that is increased.
Otherwise you can connect it just like I've shown it in the video. It's pretty straight forward so you can simply follow the guide step by step
@@jannismoore you mean to say that i need to upgrade my openAI account to paid or billing cycle to run this?
@@vishwasdhabhai229 Not necessarily. I haven't had that issue so far. It's more of a guess that is probably faster than waiting for an answer of the OpenAI support.
Do people do cold calls in Germany? Or is it illegal? I’ve been trying to whole sell in the states but in Germany the laws are way different.
I'm not a legal adviser, but I assume that if you have gathered their permission in some way (marketing sign ups within your T&C), it should be ok. Calling someone completely out of context is certainly not a thing you want to do
Yes it is only public info like corporations is allowed
@@EdemirLandau Do you have some sources on that? Would be great to share them for the community
@@EdemirLandau You have to be careful which company information you use. Anything in the Impressum isn't allowed, as that isn't made public voluntarily, but as a legal requirement.
I have copied the make senerio.. And also followed the video exactly.. But not receiving a call.. Any idea what I am doing wrong?
Do you have the phone number setup ?
@@iainhmunro Of course lol. I put the phone Id into the work flow. Someone keeps telling me I have to hard code +1 in the workflow.. Not sure what that means.. Ill keep trying
@@iainhmunro I finally got it. I had to put in the exact Google sheet id as it was pulling from a different one than I selected. Thanks for reaching out
@@TerrellLloyd27 Did you get the whole thing to run - both automations as I fail on the last Google sheet on the second automation
@@iainhmunro Yes I did. Put in the Google Sheet Id. (From the URL) on all Google sheet work flows. That should do it.
Hi Im from Sweden and When i try to Buy a swedish Phone number in Twilio i cant get the voice function. Any one can help?
You might have to verify your account further. Usually European countries are very picky with the features they allow
Wish you hadn't edited the testing call so we could hear how bad or good the latency is.
The latency is realistically around 800 to 1000ms. I’m about to release more long form content about inbound AI Agents with Vapi and I’ll make sure my editor won’t cut any potential pauses
@@jannismoore thanks, i haven't tested myself yet just want to see if it is a viable option as it is now for customer/lead interactions.
@@SlaiyAi It definitely is. We've build not only AI Agents on top of it, but whole SaaS tools. Vapi is my go to solution and you can adjust things depending on what's important to you. You can lower the delay more effectively than with platforms like bland.ai. If you have specific questions, feel free to ask them at any time.
Do you sell this as a service?
Yes - you’re most welcome to reach out: integraticus.com
Hi Jannis. Love this video thank you for sharing - 400: INVALID_ARGUMENT - Unable to parse range: 'Sheet1'!undefined2:undefined1000000 Can you advise what the issue here is?
On google sheet part of the sequence error
It seems your row is undefined. Maybe the Google Sheet you copied has been altered. I suggest going into the make.com scenario Google Sheet integration and see what comes through when you run it once.
@@jannismoore Thank you for getting back to me just earned a subscriber! if I wanted to change the copy on the Vapi and one section in the script instead of sell do you want to be a partner would that have an effect? If so where would I go to update that?
@@AyoG00 I'm a bit flushed with calls these days, but happy to answer any questions you may have. Feel free to ask here or via our Discord.
Hey, just so you know Cold calling using Ai is Illegal in the US. The business doing the call needs to have explicit consent to call the individual using a system like this.
Having permission makes it not illegal, just like you’re saying it yourself.
I think it’s obvious that you cannot just call random numbers without having any sort of permission.
If you’ve a proper privacy policy and T&C in place, I don’t see why this should be a problem.
Simply allow users to unsubscribe from cold calling, just like you’d do it with emails.
I’m obviously not a legal advisor, but this is how I personally address this.
There’s probably as well a differentiation between corporate and personal numbers.
You have to remember that the US is not the only country in the world. If it is illegal, don't do it.
@@jannismoore if you had some "sort of permission" then it wouldnt be a cold call. AI caller cold calls are in fact illegal in the U.S.
@@OrozcoJr. It isnt a cold call its a mid funnel call or mifu for short
It's not illegal for B2B in the US I believe.
Just B2C
is this free????
Well, apart from Vapi (and maybe make), yes.
If people don't even listen to other humans what makes you think they will sit around listening to robo calls lol
It's okay for many not to listen - it's the few that do want it that make the difference. :)
(Given it's actually an ethical call :D)
Just ignore ppl.witj doubts lol
You're lying, when you click on your link nothing happens. 😂
What link?
seems to be a bogus system, not a single comment says it works but lots that say it does not. Just accumulating comments for the you tube algorithm
I can make the best tutorial, provide predefined templates AND show it on the screen, and people still manage to mess it up. It’s just the nature of things, especially if someone tries to do things that they haven’t done before.
It seems you’re in the same position, so going through the comments and sharing nothing constructive is a clear representation of the position you’re in.
I read comments, so with that you just lost your option of actually getting a helpful answer, but instead a lecture to something that isn’t even partially important
@@jannismoore that is fair of you to say, so instead i decided i will actually give it a shot and not be deconstructive to you and your channel. Now having said that i believe I am at a roadblock where at the module does not work. I believe it may be to do with the the part of your video at 8:21 and it says to enter a URL into the webhook. What URL is this?
@@ammanrawjirealestate well done.
It talks about the webhook URL of the initial module you see within the Make.com scenario.
When you click on the red webhook module, you can generate a URL (You need to do this anyways before you can save the scenario).
Now this URL can be added within your account as an account-wide URL which is then used to receive any of your server messages and assistant requests (for transient assistants).
Now initially this was the only way of doing it, but now you can also assign a webhook URL (also called serverUrl) to a phone number directly. Here's the direct link for that (you'd need to update this via an API request: docs.vapi.ai/api-reference/phone-numbers/update#request.body.serverUrl)
If you're okay having it account wide, you can continue the tutorial as you see in the video while adding the webhook URL from the webhook module of the scenario into the Server URL field located within your Vapi dashboard under Settings > Server URL
Please dont edit the testing of the ai voice caller, so we see that its slow and actually not gonna work for real estate properties businesses
I’ve informed my editor already. If you consider it slow right from the start, I assume it’s more of a mindset issue than the actual technology.
Speed always depends on the right prompting, LLM, and other transcriber services. Vapi is benchmarking it’s LLMs directly during a call so it can choose the fallback model that is fastest, giving you an advantage over other providers.
At this point in time, building voice agents is always connected with a trade-off which requires you to balance speed and quality, but from what I’ve seen so far, you can do great with the right prompting and setup.
Each technology provider is evolving rapidly and bringing in new features that make the conversation more human-like. Look for example at back-channeling or static function messages.
It's actually not slow at all. If you use models such as Groq for instance and the correct transcriber etc the responses are pretty much instantaneous. Also bare in mind if you have ever spoke to anyone from a call centre or a cold call, every human will pause between answers. Usually as they are writing something on their system as they go, or running through a script so there is always a pause. These services are actually much faster I would say.
Hi @jennis moore …..it was an amazing video but I still have a doubt on how to buy a number or get it from twilio. Please explain the process of numbrr a little bit. If anyone of you know thud please help me
I would like to be onboarded in your hotel solution @jannis Moore
It'll be soon ready :)
Hello, Can you please tell me how to add new questions and it should display on google sheet like you added two columns want to sell and more details about property
You can do that by adding them via the Structured data inside of Vapis analytics tab. You can then add those fields within make.com to the required columns