Timestamps for anyone trying to follow along 0:00 Introduction 1:59 Setting up subscriptions 3:58 Creating the custom gpt 7:40 Adding webhooks in Make 10:07 Creating the gpt schema 14:12 Testing the webhook trigger 16:22 Connecting to airtable from Make 21:13 Testing gpt with airtable 22:09 Troubleshooting gpt to airtable connection 24:20 Demonstration of personal assistant 26:20 Adding additional webhooks
thank you but It didn't work for me at the webhook step :( could not connect the gpt to make. The webhook is waiting for data and gpt couldn't find the webhook.
Hello, thank you very much for your explanation. I tried to open the URL you provided, but unfortunately, I cannot access the web page. Did you change the URL or is there some problem that prevents access? Please let me know, thank you very much
Thank you so much Jamie. I've been trying to use AI for the past few months, but was unable to use it in a practical manner. Within hours I was able to make new webhooks for other tasks. Thank you again!
This looks great and love what you are doing. It really aligns with what I want to do in my Business. I have registered to be apart of the community, fingers crossed I make it in
I'm sorry, I tried following your instructions,. but at 10:08 you are jumping to the resource section without showing how you got there, what is it and where is the OpenAPI Schema?
hello! apply to the community and I will keep an eye out for your application! All of the resources are saved in there... see link in the description of the video!
Question... As oppose to creating multiple make scenarios which would require multiple webhooks, could you create (and continue to build) a single scenario with a router that would determine which branch of the scenario to run based on an initial prompt or conversation starter? I find that as I use make more and more I've got tons of scenario and having them organized becomes more and more challenging.
You could certainly try it that way and would be interested to hear how it goes! I went with the individual scenarios because the OpenAPI schema allows you to give good explanation as to what action to use in what situation.. so that your assistant will always choose the right scenario for each task!
I now have your same assistant -- now i am going to create a webhook for getting data from a chat that represents a youtube video idea and chatgpt should create draft script, etc. for me
Thanks for uploading. Some good info there but quite uncomfortable watching videos that frequently switch between dark mode and retina-searing white screens.
Yep, I have now updated the OpenAPI schema GPT... so if you generate a new schema through it, it should allow you configure the "always allow" action option now in your personal assistant!
Hello Jamie, I have created Alfred (Personal Assistant) and modified the Post within ChatGPT to be generic based on a command process that ChatGPT sets according to my request, this way I only need one Scenario. This all works great! However, when I request to Post at the completion of the processing, I get an error: "It looks like there was an issue with posting the expense. The system couldn't find the webhook." I have tested the webhook within Make outside of ChatGPT, and it works fine. Do you have any ideas about what I might be doing wrong? Thank you, Jim
Firstly, congratulations on the new addition. Secondly, excellent content as usual. I love how your videos are always so practical towards real world use cases. It's clear that the videos you make are always the summation of the cool experiments you've been working through for your own needs and interests, instead of, 'I know, I'll make video on X because it should get views'. Basically, your content is genuine and genuinely helpful and useful for that reason 😊 I look forward to playing with this when I get back to my desk! Cheers again!
This sounds very useful to learn practical ways to harness chatpgt. I applied for the community to get access to the resources. Thanks in advance for approving.
Great job! Please approve the membership request if possible. I did make my own schema though your gpt may be better to do so. It works, but I cant seem to find a way at this level to whitelist the function call without confirmation. is there a way yet? How many have you linked in so far?
Has anyone gotten access and the ability to try this? It would probably be better to have the resources instantly accessible rather than viewing the video, not having access to the tools, having to wait on an approval. By then, the interest can be lost. I see lots of people in the comments saying, please approve my request. They want to try this and it looks cool, but with such a long wait, I'm sure most won't move forward if/when they finally get access.
This was a great video and provided one of those ah-ha moments for me! Thank you Jamie. On a side note, do you know of a way to extract metadata from an uploaded image? Apart from my day job I serve as a local official and I'm constantly looking for ways to deploy workflows that help residents. In this case I'm hoping to upload an image (likely of a pothole or burnt out street light) and to then upload the image, resident provided info, and location metadata to coda/google/airtable.
Very nice walkthrough, thank you. I use Make a lot, but why bother using it at all and not just talk directly to the AirTable API and whatever other APIs needed?
Good point, I guess each API has its own quirks and requirements, and make.com or zapier or any no-code automation platform like that takes alot of the complexity (including OAuth etc.) out of the way! Particularly if stringing several APIs together...
Hi Zu! Thanks for watching... the resources for download are saved in the commuity.. I don''t see your application to the community, if you send it again I'll keep an eye out for it, thanks!
There seems to be a huge gap between ai automation tools, and what I want; even though what I want would not be difficult to create (I may have to write it myself). I don't want automation of APIs. I want an assistant that will see the buttons, text fields, etc. on my screen and click and type. I don't want to learn APIs; I don't want my assistant to learn APIs I want my assistant to do the same things I do, the same way that I do them.
@@no-code On one hand, the "operate the apps the way a human does" approach seems obvious; on the other hand, it would change the world in ways that might not be good. Today, ai uses APIs, which we know for sure are operated by bots. If tomorrow, ai uses human UIs, then I doubt there would be any way to tell the difference between bots and humans. Even for a paid account, the human can pay one time and then the bot can use the account forever after. Maybe there is a REASON there have been no good attempts at this. --shrug--
Timestamps for anyone trying to follow along
0:00 Introduction
1:59 Setting up subscriptions
3:58 Creating the custom gpt
7:40 Adding webhooks in Make
10:07 Creating the gpt schema
14:12 Testing the webhook trigger
16:22 Connecting to airtable from Make
21:13 Testing gpt with airtable
22:09 Troubleshooting gpt to airtable connection
24:20 Demonstration of personal assistant
26:20 Adding additional webhooks
thank you but It didn't work for me at the webhook step :( could not connect the gpt to make. The webhook is waiting for data and gpt couldn't find the webhook.
Fantastic! Thanks for the amazing value video 😍
Good to see you again and thank you for the new tutorial and the resources provided.
It has been almost a year, but getting back into it! Thanks for the nice words!
It seems great!
Unfortunately I can not access to the gpt to creat a gpt schema.
Is it still available?
thanks a lot!
Extremely helpful. Thank you so much. I learned something new today. good luck with everything you do. Thank you again..
Thanks for the support, appreciate it!!
Hello, thank you very much for your explanation. I tried to open the URL you provided, but unfortunately, I cannot access the web page. Did you change the URL or is there some problem that prevents access? Please let me know, thank you very much
Really good to see you back!!
Thanks Fausto!
Very helpful. Thank you so much! Hope to get access to the community. Was hoping to get onboard with this and build a transcriber for my calls.
Thank you so much Jamie. I've been trying to use AI for the past few months, but was unable to use it in a practical manner. Within hours I was able to make new webhooks for other tasks. Thank you again!
Nice one thanks!!
This looks great and love what you are doing. It really aligns with what I want to do in my Business. I have registered to be apart of the community, fingers crossed I make it in
Nice one!
I'm sorry, I tried following your instructions,. but at 10:08 you are jumping to the resource section without showing how you got there, what is it and where is the OpenAPI Schema?
hello! apply to the community and I will keep an eye out for your application! All of the resources are saved in there... see link in the description of the video!
@@no-code u suck everything behind a stupid data wall
i gotta joing to get the rescores im good
@@no-code me2
This is incredible. Thank you so much.
Thanks!
This is fantastic!!!I just applied to be a member. I hope I get picked!
I applied for community membership - thanks in advance for approving.
See you there!
Question... As oppose to creating multiple make scenarios which would require multiple webhooks, could you create (and continue to build) a single scenario with a router that would determine which branch of the scenario to run based on an initial prompt or conversation starter? I find that as I use make more and more I've got tons of scenario and having them organized becomes more and more challenging.
You could certainly try it that way and would be interested to hear how it goes! I went with the individual scenarios because the OpenAPI schema allows you to give good explanation as to what action to use in what situation.. so that your assistant will always choose the right scenario for each task!
I now have your same assistant -- now i am going to create a webhook for getting data from a chat that represents a youtube video idea and chatgpt should create draft script, etc. for me
Interested to hear how you get on!
Any updates?
I have been looking for this solution ever since ChatGPT was released. Thank you Jamie!
Nice one Ciaran!!!
Thanks for uploading. Some good info there but quite uncomfortable watching videos that frequently switch between dark mode and retina-searing white screens.
I struggled at first but it finally worked
Can you please keep an eye out for mine as well? I’ve made mistake and I’d like to see the schema as it is supposed to be.
Think you have joined, any issues let me know!
Thank you very much Jamie; Very useful Video.
Is there a way to get rid of the need for you to click the "Confirm" button (15:06) so it would work better with voice chat?
Yep, I have now updated the OpenAPI schema GPT... so if you generate a new schema through it, it should allow you configure the "always allow" action option now in your personal assistant!
Hello Jamie,
I have created Alfred (Personal Assistant) and modified the Post within ChatGPT to be generic based on a command process that ChatGPT sets according to my request, this way I only need one Scenario. This all works great! However, when I request to Post at the completion of the processing, I get an error: "It looks like there was an issue with posting the expense. The system couldn't find the webhook."
I have tested the webhook within Make outside of ChatGPT, and it works fine. Do you have any ideas about what I might be doing wrong?
Thank you,
Jim
Found it, thanks a bunch...
How long is the approval process?
Thanks for sharing this - very cool seeing how a GPT can connect to Make and Airtable.
Thanks David!
Would this builder work for N8N as well??
Extremely helpful for me too - going to get make and start using it today
😂Thanks Keith, interested to hear what you build!
Excellent video - I've applied for access to the website and I'm looking forwrad to being able to join so I can complete this work.
Nice one Ray!
Firstly, congratulations on the new addition. Secondly, excellent content as usual. I love how your videos are always so practical towards real world use cases.
It's clear that the videos you make are always the summation of the cool experiments you've been working through for your own needs and interests, instead of, 'I know, I'll make video on X because it should get views'. Basically, your content is genuine and genuinely helpful and useful for that reason 😊
I look forward to playing with this when I get back to my desk! Cheers again!
Wow thanks Richard, appreciate it! Interested to hear what you make with it!
This sounds very useful to learn practical ways to harness chatpgt. I applied for the community to get access to the resources. Thanks in advance for approving.
Looking forward to see what you offer! I have applied.
Great job! Please approve the membership request if possible. I did make my own schema though your gpt may be better to do so. It works, but I cant seem to find a way at this level to whitelist the function call without confirmation. is there a way yet? How many have you linked in so far?
Nice one! I've linked about 20 so far... what about you?!
Has anyone gotten access and the ability to try this? It would probably be better to have the resources instantly accessible rather than viewing the video, not having access to the tools, having to wait on an approval. By then, the interest can be lost. I see lots of people in the comments saying, please approve my request. They want to try this and it looks cool, but with such a long wait, I'm sure most won't move forward if/when they finally get access.
Thanks for this video! Hope I can access the resources, I already applied. Gonzalo, from Argentina!
Hopefully you have access by now, let me know if not!
This was a great video and provided one of those ah-ha moments for me! Thank you Jamie.
On a side note, do you know of a way to extract metadata from an uploaded image?
Apart from my day job I serve as a local official and I'm constantly looking for ways to deploy workflows that help residents.
In this case I'm hoping to upload an image (likely of a pothole or burnt out street light) and to then upload the image, resident provided info, and location metadata to coda/google/airtable.
❤ This is incredibly powerful. Thank you for the helpful walk-through.
Appreciate the feedback! thanks
Genious, thanks for sharing
Thank you!
Very nice walkthrough, thank you. I use Make a lot, but why bother using it at all and not just talk directly to the AirTable API and whatever other APIs needed?
Good point, I guess each API has its own quirks and requirements, and make.com or zapier or any no-code automation platform like that takes alot of the complexity (including OAuth etc.) out of the way! Particularly if stringing several APIs together...
hello, where is the link to the scheme please?
Nice explanation thank you
Thanks for the kind words!
My wish list to use the Apple Reminders to do list and talk to chat about the jobs and the order and as I do the job it gets marked off
is it any way not to push "Confirm" button for each message (that asks to send json action) to bot?
You will have to enable "Always allow" in the privacy settings of your GPT!
great video, thanks a lot for this :)
Nice one thanks!
Can't find the resource
same here. where is the link you created for chatgpt
Zu here. Very helpful stuff. I tried to download from the website, but I couldn’t find it. I did however apply to be a part of the community.
Cheers.
Hi Zu! Thanks for watching... the resources for download are saved in the commuity.. I don''t see your application to the community, if you send it again I'll keep an eye out for it, thanks!
I hope you have an affiliate link in here
There seems to be a huge gap between ai automation tools, and what I want; even though what I want would not be difficult to create (I may have to write it myself). I don't want automation of APIs. I want an assistant that will see the buttons, text fields, etc. on my screen and click and type. I don't want to learn APIs; I don't want my assistant to learn APIs I want my assistant to do the same things I do, the same way that I do them.
I think there are already prototypes of what you are looking for!
@@no-code On one hand, the "operate the apps the way a human does" approach seems obvious; on the other hand, it would change the world in ways that might not be good. Today, ai uses APIs, which we know for sure are operated by bots. If tomorrow, ai uses human UIs, then I doubt there would be any way to tell the difference between bots and humans. Even for a paid account, the human can pay one time and then the bot can use the account forever after. Maybe there is a REASON there have been no good attempts at this. --shrug--
Holy Shit this is amazing
😂thanks Joe!
find most american talks are builltshit to show their presumption.