Lesson 10 - ChatGPT Blocks in Zapier: Outputs [PART 1]
Start an AI Automation Business📅 2024-07-31
Transcript
[00:00] welcome to the third module in this course where we're going to be looking at pragmatic skills you can start using in your AI automation agency in this lesson we're going to be looking at everything we need to know about the chat gbt block and the open AI block when it comes to formatting more specifically we're going to learn how to do large data outputs and large data inputs we're going to be learning different methods and ways to ensure consistent outputs and a ton of other stuff let's go ahead and jump in all right so I went ahead and opened up a zap your flow here I'm going to use a
[00:30] random trigger we're going to use a scheduler because of the fact that in this tutorial we're just going to be looking at everything we need to know about these specific blocks and their functionalities to start off let's understand the fundamental difference between when you would use an open AI block compared to a chat gbt block the first thing I want to point out here is that even though open AI has the event of send prompt never do a chat gbt prompt within the open AI block always do it within the native chat gbt block as it's going to allow for better formatting and better outputs went ahead and changed how this looks right now so
[01:00] it's a little bit easier to digest so I went to the visual editor as you see here so let's go ahead and continue now that we know that when we use an open AI block never to use the prompt function what can we do with the open AI Block open AI block is going to serve four major functions it's going to be able to do transcriptions translations generate images and API request from our experience here at webcafe Ai and the clients we've worked with the only two main ones that we've used constantly or generate image and generate transcription let's go ahead and
[01:32] understand the event of create transcription and everything we need to know about it let's go ahead and select it we're to continue here we're going to choose our open AI account so now that we've done the event of create transcription we're going to have a couple of different action inputs that we have readily available to us such as file prompt response format and language of audio now as you'll notice for the file type it does give different formats but as a rule of thumb always use an audio file due to the fact that if you were to upload a 10-minute video compared to a 10-minute audio file the
[02:02] size of the video is going to be much larger and it's going to take a lot longer for open AI to go ahead and digest it for the transcription to an audio file which file size can be tremendously smaller than when you compare a 10-minute video next what you'll notice is that you need to provide a public URL of the file that you want the transcription for let's go ahead and just show a real quick way to do that so this can help you with other versions of API you may to do in the future to do so go ahead and create a Google drive folder and we're going to go ahead and make sure that the
[02:33] underlying folder we press share that anyone with the link can essentially access it this is the first step next as you'll notice we have our audio file that we're going to do a transcription for so let's go ahead and just copy the name here add a step here and we're going to do a Google Drive step now this obviously can work with others you know cloud-based storage such as Dropbox we're going to do an event of find a file and continue here here and we're going to go with our courses account here continue put our file name so in
[03:07] our context it was gbt audio so test this action as you see here it shows up perfect we come over here and we can go ahead and find the file that we just provided here you don't want to do the title you want to go ahead and do the file exists but not shown and now we have our file and one thing you'll notice which is very important is that the it's a M4 a smaller file size and it's not going to be an mp4 which would be a video in that context now that we know that I'm going to go ahead and delete this step let's go ahead and jump further here so one thing that we notice
[03:39] here at webcafe Ai and this might come up with different things that you're doing as a business or when you're working for other businesses is that certain words during transcription are always spelled wrong no matter how clear you say them over audio now for us an example of that would be quite literally chat gbt anytime we say the word chat gbt with audio and have to deal with transcription it spells out to either CH h a d gdt gbt it never really spells out to chat gbt so the use case for prompt
[04:10] essentially is that you're going to be able to identify certain things that happen in your transcriptions and make sure to correct them so in this context what we can do is anytime that chat gbt is said make sure to spell it like we do semicolon we do parenthesis here and then we'll do the correct spelling which is going to be chat gbt now you might be asking yourself well Corbin you just said that in the audio
[04:40] files they don't actually spell it right so how would it know that I'm referring to chat gbt with the correct spelling well this is smart enough where it would be able to interpret the CH h a d gbt the misspellings as close enough to chat gbt that would be able to reformat them to the correct output which would be you know the correct spelling chat gbt next we can choose our response format here text SRT W or webvtt Json uh verbose Json you know I know personally
[05:11] uh SRT is a very valuable format when it comes to transcriptions so that's pretty cool and finally you're going to want to put the language of the audio and the way that they want you to format it is called ISO so you put that would be for English you can click this link and then it'll give you all the different formattings depending on your language okay great that is create transcription the go ahead and check out create translation same concept here essentially what you would do is you would upload a file such as a audio file as you see here these are the acceptable file types that can be uploaded here you
[05:42] would put a similar type of prompt and then you'd be able to choose the output as well all right let's go ahead and check out the next feature here we got generate image now I really like this one as this allows for seamless content creation when you really know how to effectively structure your prompts when it comes to delay that is the system they use for AI image generation right off the bat always upscale it to 1024 x 1024 unless you're testing prompts as you're going to want the highest quality possible when generating an AI image this essentially makes it so you
[06:14] can generate variations of the image I wouldn't suggest this in the context of mass production maybe use this in the context of you're testing a prompt and you want to see multiple variations so you'd put you know either five or six here I would say there are two main fundamental things to understand when creating a delay prompt first is the resource of prompt base so essentially prompt base is a marketplace where individuals can sell AI prompts here all you would need to do is go to Marketplace select delay and then there is a plethora of different types of
[06:45] pre-made prompts using delay engine you can fine-tune your specific category over here so if I just wanted 3D I can click this so understanding that essentially you'll be able to choose any type of design you want to in whatever context you may be generating an AI image for so for this one if I like really like this I can see examples and essentially post purchase it'll provide a prompt and you will essentially input a variable so let me go ahead and show that so this is one of the prompts we sell here on prompt base and it's called Pablo Picasso art once you purchase a
[07:16] prompt essentially you will be provided with the prompt right here so let's go and copy that and you'll be provided with a testing prompt for so an example variable that could be inputed there and from there you'll be able to see you know different types of design examples one over here went ahead and pasted that from prompt base so here's number two of what's important of understanding here is the actual underlying variable that needs to be inputed go ahead and use live data from zapier so what do I mean by that essentially anything that happened previously in the step can be inputed here so this doesn't really make
[07:47] sense but I could put Monday here or the date and essentially that will make sure that every single image generated will be a new type of image now that being said in theory I could put dog here and if I generated this a 100 times every single one of those images would still be different the question is whether you want to have you know a really a different topic but use the same style image if that makes sense