Chat GPT is amazing when it comes to automating tasks, but only if you know how to use it right. Oftentimes people are left copying and pasting stuff into chat GPT and outside of chat GPT, and ultimately that wastes time. So in this video, I'm going to show you how you can stay in your GPT and pull in relevant information, whether it be from WordPress, Google Docs, Google Sheets, social media posts, emails, whatever it is, you have the ability to stay in your GPT, click a button.
and retrieve that information in seconds. You also have the ability to send that information. Maybe you want to send an email. Maybe you want to update a row in your Google Sheets from your GPT without leaving your screen. Maybe you want to create a Google Doc, analyze a PDF, whatever it is.
You have the ability to automate that and stay within your GPT the entire time. And in this video, I'm going to show you how it's done. I'm going to be giving you so many resources.
I'm going to be giving you a step-by-step checklist. And I'm just going to ultimately walk you through how to do this. and how to see success when automating tasks in ChatGPT.
Listen, AI should be saving you hours of your time right now. If it's not, then you're doing something wrong. That's why the AI Foundations community was created.
It was created in order to enable you to be more efficient and more productive, whether that be tasks in your personal life or tasks in your work life. We show you the steps to get there. If you want to learn how to leverage AI in your industry to become more efficient and more productive as a person, then join this community using the link in the description or the top pinned comment.
You will not regret it. Before we get into ChatGPT and we start learning how to actually automate things, let's understand real quick how it's working under the hood and some of the terminology I'm going to be using. So the red line is representing getting information from an external application. The blue line is representing performing actions on other applications. Okay, so you start off in your GPT.
This is Big Chat GPT logo. And you can either send a get or a post request. A get request is the red line because you're getting information from other sources. The post request is the blue line because you're posting information on other sources or to other sources. So let me show you a quick example of a get request.
I might prompt to my GPT how many new emails do I have today. And it will send that request to Make and run through an automated Gmail process on Make because it connects to Gmail and fetch my email contents and how many emails I have and the body content of those, etc. And it might form back a response of you have five new emails. An example of a post request might be me saying, write a story about a dog and send it to my Google Docs when complete. My GPT would take that request in the form of a post request, which I'll show you how to set up, contact Make with the content that I've provided, the story about the dog, and send it to a new document within my Google database.
And what I've even provided for you is a step-by-step checklist down in the description that you can just follow along as we're doing this video together. for everything you need. I've even provided you a custom GPT for this system to help you when I'm not here to help on the video. So first you need to understand what you want automated, and I'm sure you have some ideas right now, even just by watching this first part of the video. But let's say for me, I don't have much time to create LinkedIn posts, but I read a lot of artificial intelligence news articles and I have specific websites I like going to.
So for me, a good example of creating a GPT. that is completely automated and posts on LinkedIn for me would be step number one, using a get request so I can aggregate all of my news articles and bring them back into chat GPT so that chat GPT can automatically create LinkedIn posts for me based on the relevant content from the news feeds that I like. So I'm pulling articles into my GPT, creating posts using chat GPT.
That's why we would go to chat GPT for this. And then I would send off those posts to my LinkedIn automatically. That's what we're going to be building today. So let's get right into it.
Step number one, what we need to do, as you can see on our GPT automation checklist is we need to create our GPT. So we need a name, a description, a profile picture, instructions, and conversation starters. If you want a more in-depth guide on creating this version of your GPT, like creating a very good GPT with amazing instructions, I'll leave a link to that video in the upper right hand corner where I go much more in-depth on that.
So real quickly, I'm going to create this GPT and then I'll report back to you once I'm done and kind of go through what I did. What I'm going to do is in the upper right hand corner, I'm going to go to my GPTs and then I'm going to select create a GPT. Here I can give it a name, description, instructions, conversation starters, and a profile picture. And again, since I am automating my LinkedIn posts, I can give it a name of the LinkedIn master and then I can give a little description of what it does. This is actually an image of the LinkedIn logo that I got generated for this on.
grok which is another language model i can hit open and just like that we have a profile picture name description and now we need instructions and this is kind of going to be instructions for how your gpt needs to run all of the actions you want it to complete and how what's going to contact your webhooks. So for that, I put you are a LinkedIn growth specialist. Now this is basics of prompt engineering.
If you don't know this yet, then you definitely need to learn it before you get into fancy automations. But I basically give it a role and tell it its job. I tell it what it's going to receive.
And then I say right here, when I say fetch articles, contact my webhook and get me the most recent articles on AI. This is going to be a send-off command to start the get request, which is going to be important. As you can see, it says when I say fetch articles, in order to make my job easier, I can copy fetch articles and make a conversation starter.
So now I can hit this button and say fetch articles, and it will automatically run through that process for me since I set it up like this in my instructions. And then I basically tell it to receive the URL, read the content and create a LinkedIn post. And I give it a custom output format for each LinkedIn post.
Now I can go back to our GPT automation checklist. And step one is complete. We have a name, description, profile picture, instructions, and conversation starters. Next, this is where the real fun begins.
We need to create the connection in make.com by creating a webhook and creating our make.com automation. And a lot of the times it'll take 10 or 20 minutes to create an automation, but it will save you hours of time on the backend. And I even give you a GPT to walk you through this step-by-step.
And it's right here in a link. So it's called the Schema Ninja. And I created this just for this video. And what you have the ability to do is hit get request or post request, and that will actually send off a set of instructions to the Schema Ninja in order to help you create that.
So if we go back to my instructions, what you'll notice right now is I need to first set up a get request because right now I say, when I say fetch articles, get me the most recent articles on AI. So how am I going to do that? Because I need to get that information back into ChatGPT, therefore to get request.
So I can go to my Schema Ninja and select to get request. And it's going to walk you through step by step what you need to do in order to create this request. As you can see, step one, create a webhook and make, and then change it to these settings right here. I even give example images for each step so you can see what I'm talking about as you're using this GPT.
But keep in mind, I'm using a free make account for this, okay? I'm not using anything paid other than my GPT Plus plan, which I'm sure if you're watching this, you might have. Right now, what we want to do is we want to select create a new scenario. and then follow the instructions from the schema ninja.
So I need to create a web hook and change my settings to get request headers yes and get HTTP method yes. So I can do that by once I create a new scenario, this box will pop up, I can type in web hooks, and then I want to select custom web hook. And here you need to add a web hook. So I'm going to add you can give it any name you want.
I'm going to name it posting for LinkedIn. Beautiful. I'll hit show advanced settings. And turn on these top two and then hit save.
Then I can hit stop and OK. Beautiful. So now we have the ability to contact this specific URL in order to start an automation that follows this.
We can start an automation by adding modules here and connecting to so many different apps like Facebook, for example, or Instagram. And we have the ability to download media, get media. We could contact Notion and create database items. We could contact. Google Sheets, for example, and we can create rows, we can create sheets, create sheets from a template.
So as you can see, there's multiple different things you can do after this webhook URL is contacted with our GPT, which I'm going to show you how to set that up now. So you can just keep walking through this. We've created our webhook in our settings. So I'll just say complete. And it's going to keep giving you step by step instructions for what you need to do.
So now it says create your automation using make modules. So the tool I use for aggregating my news articles is a tool called rss.app. This allows me to create feeds and connect with information from multiple different sources.
I have articles pulling in from MIT News, Artificial Intelligence News, OpenAI's Research, and TechCrunch. So this URL aggregates every new post that comes from those websites. I can actually connect to that and make.
So once this webhook is triggered, I can type in rss, and then I can retrieve feed items from a specific feed that I create. So I can put in my URL and then I can have a nice custom date from and date to. Most of you won't be needing this complex of a format for date from and date to, but I just want to set this up so I can be receiving information from yesterday. Maximum number of returned items. How many articles do I want coming in at once?
Maybe I only want five. I can hit OK. And now I need to format this data so that it's ready to be put in a webhook response.
So I have this set up. I'm pulling in the URL, title and author. I can hit okay and now I can aggregate this data in order to pull in each article because right now it would only pull in one article so I have to make one text strand for all of them.
So I did that that was very very simple in the text aggregator side of things and many of you won't have to do this complex of tools in order to aggregate some information. But next after we've created our automation we can take a look at everything we've done. So we've created our web hook and now we need to finish creating our automation.
So we can do that. by just talking to our schema and just some more. I'm just going to say complete right there. And now the next step after we have our automation done, which is these three steps right here, what we need to do now is create a webhook response and map our data.
So the webhook response is important because this is what your GPT is actually going to receive. So if I go here and I select webhooks again, and I select webhook response, in this body field, this is what information is going to be given to your GPT. So I've set up this information in a way to where all I need to do is get my text from my array aggregator.
And if I have five articles, then I need to make five responses of that array aggregator. in the body of that webhook response. But you can just put normal text in here too.
Maybe you just want to put, hey, with a couple of exclamation points, this response will be received back in your GPT. But when you're using these applications, you actually have the ability to map specific data to your webhook response. And this is where the dynamic ability of your automation comes in play here. So I can hit okay, beautiful.
Now we have this automation pretty much all up and running. We've created our webhook response, just as the schema ninja has told us. We have an example image here too. Now I can just say complete.
What do we do next? Now we need to create the schema for this get request. So we can actually use chat GPT to pull in the information to our GPT here, which in this case is the LinkedIn master.
So if I go back to the checklist, actually, what we've done is we've created our automation and now we're on the step three, which is creating our GPT actions. We need to go back into our GPT that we've created. and actually find a way to connect to this automation here. So the schema ninja told us to create open AI schema for the get request.
Take a screenshot of your webhook response body and upload it here, or you can just give it the information it needs in order to create that schema so that it knows which information to pull. For now, I can just go to my tools here, and I can paste in a screenshot of this information that I want ChatGPT to be pulling in, URL, title, and author. So we need to take a screenshot of the information we want pulled in, or we can just tell it. Maybe I want article URL, article author, article content, whatever it may be.
You can give it all that information that it needs in order to create the schema. Then you can tell it the goal, which one and two kind of go together. This is just to get an idea of how to create your schema for the get request. And step number three, it tells you to provide it with your make webhook URL that you created.
in the first module right here. So this URL, this blue URL under your webhook action. So I'm just going to keep following this step-by-step process.
I'm going to upload my screenshot of the things that I wanted pulled in, or again, you could just tell it the things you want pulled in as well. I just want to show you an example of this multimodality. And then what you want to do is copy your webhook URL, the entire thing and paste it in. So I can say I've provided everything, create my schema and send it off.
and it's automatically going to create your schema for you in the JSON format that you need. This is what you're going to paste in to the action section of your GPT. This is that long string of text that is always very intimidating whenever you want to automate something. So if we go back to our GPT now and we scroll all the way down to the bottom where it says actions, down here we have the ability to hit create new action.
Here we need to upload the schema that was just created. by the schema ninja based on all the properties you want pulled in from your automation. So I can copy this code, head back to chat GPT and paste in the schema. Just like that, we now have a get request with our specific path, our operation ID, and it's very beautiful.
So now we can even test this. But before we test it, we can just go back to our schema ninja. And we can say complete now that we've uploaded this, there was no errors, nothing went wrong. We have a beautiful action in here ready to go. And this right here, when we hit test, is going to contact this web hook, which will trigger the automation we just created.
And after it runs through this three-step process in the middle of the red modules, what it will do is send back what it got from running through that process. Okay. So I'm going to turn this on in the bottom left-hand corner, and then I'm going to hit save on my automation.
Because once you hit complete in your schema ninja, that's what it's going to tell you to do. is to save and test your information. Then we can send off a test. So I'm going to hit test. And what it should do is pull in all the information that we're getting from my RSS feed, which is an external app that I wanted connected.
So I'm going to hit always allow up here. And it's going to talk to this webhook. It's giving me the posts now from my RSS feed.
So this information is very custom because it's coming from my specific feed that I wanted in that external app. And it's actually making LinkedIn posts out of these right now. So the data came in, but maybe it didn't come in in the format you want, or maybe it didn't come in in the exact way you wanted. As long as you get response received with a status code of 200, then it means your automation is working.
You have the ability to change whatever comes after this webhook, update your schema that you've created with Schema Ninja, and actually pull in whatever information you'd like based on this webhook URL. As you can see though, it's just giving me the post and it's not showing me the title, author, and URL first. Maybe first I want to actually read the article, then tell it to create a post based around it. So I can change the instructions and edit it around at this point, but we've completed our get request.
So as you can see, I've added in some very basic instruction updates. I say, before you create LinkedIn posts, provide me with the following author, author here, title, title here, URL, URL here. Then ask me which article would you like to create? to create a LinkedIn post around.
That way I have a little bit more control and it's a little bit more of a process rather than just creating articles and posts right from the jump. So now I can go back and I can test this again by hitting fetch articles. Now that we have our get request set up, since we set up the instructions to contact the get request once we say fetch articles, our conversation starters will be working.
So I can hit fetch articles and test this out again to see if it's pulling in data how I want it to. So as you can see now it's coming back with everything I asked for. It's giving me the author and then the title. And the title is actually in the form of a hyperlink which is beautiful. So now I can actually click on these articles and go to the source where they're coming from.
And then it asked me, which article would you like to create a LinkedIn post around? I could just type out a number at this point. I could just say number three, for instance.
Then it's actually going to create that post around the article. So it's starting off with, what if we could design proteins to revolutionize medicine and beyond going a little bit in depth and then having a CTA. And then it asks me the question, would you like me to send this as your LinkedIn post?
I could say yes, but we don't have that action set up yet. But we go back to our checklist. We now have the ability to check off the remaining things because We've created our GPT action and we've tested our GPT action.
We've grabbed test data from the get request. Now, what if we actually want to send these LinkedIn posts? Well, then we need to revise our schema because our schema right now, if we go to our actions, only has one method of using the HTTP request, which is get right now. So we only have one available action right here, and that's to get webhook data, or in other words, get our feeds and our news articles from that app. What if we wanted to send this?
Well then we would need to add to our automation and we'd also need to revise that schema in order to make it available to have another method of HTTP requests. And in order to edit this, first we need to have test data with our webhook so we can dynamically map fields. And the webhook data needs to be the LinkedIn posts.
We need to figure out a way to pull in dynamic LinkedIn posts to this webhook. So we can do that very simply. First, we can go to our schema ninja GPT and type in a new prompt and make sure you're staying in the same chat thread for this.
So what I'm doing here is I'm basically staying in the same thread because we know that this schema is working and this GPT is designed to create schema that works. So what I can do very simply is add in another HTTP method to the schema that's already working and it already exists. And then later we can change around our automation after we send it some test data.
But I say, using this same schema, add in a post request with the get request. I want this post request to be able to send my content to the webhook. The content in this case would be a LinkedIn post. Help me revise the webhook in order to be able to add another HTTP method of post where I can send information to my webhook in the form of a post. I keep reiterating in the form of a post.
So. At this step, if you want to not only get information into your GPT, but be able to post it elsewhere, which some people are fine with just pulling in the information and then working with it and then doing whatever with it later. A lot of the times that's the hardest part is getting that information in, but we can actually revise our schema to add a post action to do something else.
So I can send this off. I recommend just using a similar prompt if you're trying to add to your schema to get a post request query. But what it's going to do is it's going to use the same URL.
It's going to use your same webhook path. Except underneath get it will have your entire get operation and then it's going to go into a post operation So we're going to send a LinkedIn post to our webhook Very very cool. It gives us our updated schema So right away what we can do is we can copy this updated schema that it gives us go to our GPT and actually Paste it in but first you want to delete your old schema, which might be a little scary, but then you can paste in new And what's going to happen is you're going to see another method down here. So now we have the ability with our GPT to get information and post information.
But as you can see, we have some required fields down here. It says post title and post content. This is what we want our GPT to format our information with. We want a post title and post content in order to actually send these posts to our webhook.
So now we need to go edit our instructions to ensure that whenever it creates a LinkedIn post, it formats it in this JSON format. And that actually with our schema ninja gives us that format that we need. And it gives us instructions for a post request.
So what I can do is I can just copy these instructions, head back to our GPT, and then go paste those instructions in our instructions. So I can leave these at the very bottom. And these are going to be instructions for the post request.
When using this post request, you'll need to send a JSON body that contains post title. You can put those in quotations. And. post content fields because you need a structured way to send your data to your webhook in order to be able to use that data dynamically.
So that's why we're doing it in JSON format. So underneath these instructions for post request, I can then put that example JSON body. I can say here is an example of how to structure this data. And I can go back to the schema ninja and copy this code, the example JSON body for post request. And if it didn't give this to you, just ask for this stuff.
And the schema ninja is really good at doing that for you. And then I can paste it in. Beautiful. We have the structure of how this should look.
I can then hit close. And now we can go test our post request to make sure it's actually running. So I can go to actions and under available actions where it says post, I want to test this so I can hit test on post and it will send an example post to your automation on make just so we have data to pull from when we do create the rest of this automation. I can hit confirm.
And I can kind of run through this process. I'll just say article three, and then I can say post. And when I send that off, it should perform the action of the post method.
And it says the LinkedIn post has successfully been sent. What I'm going to do now is actually map this data to our web hook and create a LinkedIn post around it, like physically create the automation that we can use from the data that we're receiving from our GPT here. So I'm going to unlink in between right here. And I want to have a router because a router is going to allow us to go to different methods. So I can select router.
And this is going to be the router to go to our get request. This will be a fallback method. And I'll explain what that means in a minute.
But now I can add the linked in And once I hit LinkedIn, I can just do create a user text post and add that to the router. So now we have two different routes to go the post request method and the get request method in order to make sure that it's filtering out correctly. What we need to do is set up a filter on our post request.
We can select the wrench, hit, set up a filter label. This can just be condition setting. And then for condition, we can have method is equal to post. and we can just make equal to case insensitive.
And then I can hit OK. So now if it is a post, it will go here first. And this will be a fallback route, we can set up this filter in between the RSS feed and the router to be a fallback route.
And this will be the first condition it sees and checks if it can go to next, what you need to do is you just need to hit send data again, this is just going to make sure that your data is being set in your test feed. And I just do this in order to make sure the fields are pulling in because sometimes When you go here and your automation isn't saved, when you try to map in your content to the dynamic fields in your webhook, they won't pull in. So what you need to do if it's not pulling in is save your automation and hit this little back arrow and then go to your GPT where you've been testing and just hit send data again. And then when you click back in here, you should be able to dynamically pull in that content.
So now I can pull in post content just like that. So this would be an example of setting up a post request. We need to be able to pull in information dynamically from our webhook that's coming from our GPT like shown.
I can hit okay. I can save this automation and now we can put this to the test. I'm going to go to my checklist, check off the post request. Again, you don't have to do a post request optional depending on your automation. And then what I can do is I can come to my GPT and now that all the testing is done in the upper right hand corner, I can hit create and then I can make access invite only.
And then I can hit update or I can create it for the first time if that's what you're doing. And then I can hit view GPT. Next, I can just walk through this process and do it. So I can hit fetch articles. Then I can hit always allow on that web hook because it needs permission to talk to it.
And that's going to give me all the recent articles from my feed that I can create posts around based on the last 24 hours. So maybe I could say post number three and send that off. And now we have this automated process to where I don't even have to leave my GPT screen.
And it's going to give me a post draft. It's going to give me this beautiful LinkedIn post based on the article. And I could say, revise it to attribute content to the author at the end of the post. You can even revise these posts and you can revise your content.
And ChatGPT is basically your assistant now. You can ask it to go get things for you. Then you can ask it to post things for you.
So I'm telling you to provide a link of the article at the bottom instead of the inspired by comment, because I don't want it to just say content inspired by. I want it to provide a link to where I am getting these. thoughts or these ideas from.
So now it asks me, would you like to send this post to the webhook for LinkedIn posting? I can say, yes, absolutely. So as you can see, I'll just have to hit confirm really quick, very easy.
And then it's going to post that on LinkedIn. Your LinkedIn post has been successfully published that used the information that it got from here. Very quick and a very automated system.
As you can see, if I go on my LinkedIn and go to posted just now, I can click this and it did post literally everything. in a matter of seconds, just like that. That's how innovative this is. And that's how efficiency works within AI.
Now, if you enjoyed this and you want to even become more efficient, then again, I recommend the AI Foundations community to anybody. I mean, since just recording this video, we already have two new members in the past hour. People are joining in order to leverage efficiency with AI and people that are applying this stuff in their business and their personal life are succeeding with it. There's so many different industries in here leveraging artificial intelligence, and you could be one of them. So if you want to learn more cool things like this to save you hours of your time using this technology of the future, then I don't want you to fall behind and I want you to join the community.
We have so much to offer there, live calls, courses, et cetera, in order to get you from A to B. But with that being said, I hope you enjoyed this automation and I hope it helped you. If it did, please subscribe and like this video.
I will be coming out with more content like this. If you'd like me to go more in depth, just let me know. I will be happy to answer your comments. record more videos, automate more things so I can help save you time with AI. All right, I'll see you in the next video.