Speaker 1: In this video, I'm going to show you how to create a 100% AI automated social media system that can take an idea or an article or a piece of news and transpose that into a highly optimized LinkedIn post, tweet, Facebook, whatever you like. And it will also create AI generated images specific to the dimensions of that platform all from one single input. The Twitter GPT, the LinkedIn GPT, they've all been trained on the latest best practice about what's actually working. And so we can take the best automation and what's actually working to create an incredible flow and all the guides, resources, the prompts for these GPTs and lots of other cool resources are all available in the school community I just created down below. And if you have any other questions about this, I'll be able to help you directly better as well. And so the objective of this video is to create this AI powered social media system that works but also delivers really good quality output because we want the sizzle and we want the substance. So we'll never cover automations that look cool but don't actually physically work. These are things that actually work and it's stuff that you would use. So what does the automation look like? We have a trigger at the beginning that could be Slack, it could be Google Doc, it could be any messaging platform that you like. That then goes through to Perplexity AI, which is this incredible AI search engine. If you're not familiar with it, it's like chat GPT, it makes Google and it can just pull and understand URLs and links. So it's pretty amazing. So in this instance here, we're going to find a piece of news or information. Let's say that we're in finance space and there's a new article about the S&P 500 or the FTSE 100. The way this automation is going to work is we give the news or the update or the prompt or the idea to Slack and all we ever do is enter a URL. That then goes to Perplexity AI. Perplexity AI will then get that, it will then summarize the entire URL for us immediately and automatically. Once that happens, we go through to this router and this is kind of where the magic comes in here. It goes to four separate custom GPT agents. We have a, as you can see, a LinkedIn bot, we have an X bot, we have a Facebook bot and we have an Instagram bot. And the reason why we've got four completely different GPT assistants is because the outputs we need for each of these social media platforms is completely different. So this isn't going to be generic. It's going to specify it specifically to the social media platform that you're posting. As you can see, the X bot has a completely different set of instructions and parameters and examples compared to the LinkedIn agent that we have. So then these bots will take the summary, turn it into an amazing post for us. Once that then happens, we're going to get Dolly3 to create a specific image for that article in the dimensions of the platform that we're going to be using. And then once it's done that, it will just post automatically to any of the platforms that we want. And if you want add specifics or add your own tone of voice, I'll show you exactly how to do that as well. And to run this automation, you only need four things. The first thing is a make.com account. I've got a link in the description down below. The second thing that you need is perplexity credits. As you can see down here, if you come onto my account, go down to the API and you can buy credits for like $5 and it will last you a very, very long period of time. Like if, for example, I've only used one cent and I've done so many automations with it. You'll also need this perplexity AI integration from make.com. And then finally, some credits from GPT. And if it sounds like a lot, don't worry, I've got you. I made a video literally four days ago covering exactly how to set all these things up in detail and create autonomous AI agents. And if you haven't seen that video yet, I encourage you to watch that one first, and I'll be waiting for you here for when you come back. And with that in mind, let's get started with this incredible automation. Right, so in the make.com dashboard, we come top right to create a new scenario. So let's think of our trigger. Now, the trigger that I'm going to use here is going to be Slack as I did in my last video. The reason I do that is because that's just the software that I use. But if you've got a better trigger, feel free to use that. When I click on private channel messages, perfect. And then we select the private channel from which we want the message to come. I've selected automate in private. And then we put the limit for how many messages we want it to check. It can be as low as high as you want, but it should always be high if you're going to put the inputs there. And then we click okay. Brilliant. Now, how often do I want this automation to run? Since we're going to be posting on social media, what we don't want to do is send off 50 posts at once because it's spam and some platforms will suppress you for that. So what I propose that you do here is run it every 15 minutes to every hour. So I'm going to put here every single hour. I want this automation to run. Excellent. But set a timeframe that you're comfortable with. And that kind of makes sense for you. So I've just Googled here finance news. Top article is Eshell trying to kill the London stock market. Let's say we think the article is interesting and it's something that our followers would appreciate. And we want to share our thoughts on what that looks like. So we've copied the URL. Only thing you ever need to do this entire automation is paste the article like that. Boom. It's done. So what we want to do is want to test this as work. So we right click run this module only. And let's see, we click on the one to show the output come down. And the text is that excellent. I always encourage you to check the automations as you go along so that you don't build 15 modules and realize it doesn't work. And the next thing we're going to add is perplexity AI. So click add new module. We're going to search for perplexity. Now, before you buy this again, you'll see the community. You click the community. You will open a new tab, but you can follow to purchase this. I don't, isn't that expensive, but the other video covers that in detail. So perplexity AI we click here. Okay. So then we click on create chat completion like this. And one quick point I will make on the trigger before I explain this is that there's a way to do this where it will scrape a webpage. So let's say you've got a news article page that you like a lot and they keep posting updates and you want to take that and I can show you how to do that in another video. But for the purposes of this one, I want it to be something that we select. And we think is cool. We got a hundred articles. That article is the one I want to cover, or that's the news update. So this is going to be user-driven, but we can completely automate that. If you want to see that, just let me know down below. So we're here, Mistral 7b instruct is fine. Click add item for the message that we want. And what we're going to say, we're just going to say, I would like you to summarize this article and the prompt here is very clear, very obvious. And then as you can see on the right-hand side here, we're just looking for the text and then we just have text. And then we want to sign the role. The role that we're going to use here is user because we're instructing it as a user. We don't need the assistant functionality. And then don't worry about these other aspects like tokens, temperature, and top P and K for the moment. Click okay. That's cool. Then we'll click run once to run the whole automation together. So Slack has found the message. We can see here, it's got the article that we're looking for. We come back over to perplexity and we come down to choices. We move it over for you so you can see. Cool. Right. We're here. So let's have a quick look. Come here to perplexity, choices, plus one, message, content. I'd be happy to help you understand the main points of the article, blah, blah, blah. The article discusses Royal Dutch Shell's potential delisting from Lindenstock exchange and its implication with UK's financial market, blah, blah, blah. So we've gotten guys and gals a full summary of an article from a URL, which is pretty cool. But this is where it gets better. The next one I'm going to add is going to be a router. So what the router basically enables us to do is to split off into separate different channels. So for the purposes of this, let's add a chat GPT, pull it up. Cool. And what we're going to look for here is to message an assistant. Great. We're going to message an assistant. You'll see this. If you haven't connected it, you'll be prompted to do so. Again, all covered in the last video. Message. Great. So what we wanted to give it now is the output from perplexity. So the way that we configure this assistant, it will already know what to do. It's all the instructions. So what we put in the message part here is you come down to choices. We get on to message and content. And that's it. That's literally it. And then we pick the assistant. The assistant that we want here is LinkedIn bot. Brilliant. Cool. Okay. I'm a little hack for you. All we're going to do for this is we're going to clone it just so it saves a little bit of time. We'll delete this one here and it automatically snaps into place. And all we do is we come here. We then sort of a LinkedIn bot. We go for the tweet bot. Let's have a look here. Tweet bot. Perfect. Now I'm going to show you the assistant and how that works. So we come back over to playground. We're going to assistants. And we've got a LinkedIn bot and our tweet bot. To create a bot, what you do is you create the top right hand corner here. You give it a name. You give it some instructions. The model we're going to be using is GPT-4 turbo. It's an excellent model and it's absolutely worth it. And it hasn't been expensive anywhere near the cost to be honest with you. And at the bottom right, you can see you can add files. Add files like turn a voice, best practices. Whatever you think is going to be helpful to optimize the model. Now, in terms of what we've actually done with these models, the tweet bot and the LinkedIn bot. I'm going to show you what it looks like here, for example. Now, what's really important to me is these automations work for you guys because I don't just want to give you automations. Oh, it looks cool. But they never show you the final result. It needs to work. So the autonomous LinkedIn agent, for example. We start with this prompt here. Act as an informed LinkedIn expert in insert industry. So whatever industry, you know, you want, you can customize it in the school community. You'll be provided with the summary of an article. You'll create an optimized LinkedIn post based on the article. Incorporate LinkedIn post in best practices and using the outline structure below. So we've got things here like provide the main context, the main idea. Start with a brief summary on my part of the article. You want to transform into LinkedIn post. This includes the key message, the article that they present in the article. Add personal or relatable anecdotes if it's appropriate real-life scenarios. Incorporate engaging elements. So questions to the audience. That's really key. Compelling statistics or data, which we'll get from perplexity. Calls to action. Relevant emojis for emphasis to draw attention to key points. And then we've got style and tone. We've got hashtags and keywords. And then we've got additional details for them. And the real magic here for this, as you'll see, is the examples. So what I've done with this LinkedIn bot is I've trained on some of the most viral, critical examples of what actually makes a LinkedIn post successful. So I've got all the LinkedIn posts that had phenomenal engagement. Example one, example two, example three, example four, example five, example six. It goes on and on and on. So we've selected all this stuff. What we're going to do is copy all of it. We're going to go back over to our bots. You paste it in like this. Come to the top. Get rid of this bit here because I just copied that. And then there you go. And then we insert the industry. So the industry that we want is going to be finance industry. Perfect. And then we've saved it. GPT-4 Turbo. That's perfect. A little hack for you guys as well. If you want to play around to see how good it is, what you can do is come on playground, go on assistance here on the left. And then what you can do is chat with your bot. So this helps you optimize it. And when I was optimizing these bots, it took ages in terms of like trial and error. And sometimes it gives you something that's like 90% there, but not quite perfect. It can be a little bit of a nightmare sometimes. So what we're going to do again now is we're going to come back over to Slack. We're going to repost the article to re-trigger the entire automation. We come back over to Perplexity and to our automations. And then we hit run once. Into meg.com. Slack identifies it. Perplexity is now beginning to summarize it. Goes to the router. And there you go. It's gone to OpenAI to create the LinkedIn post. And this one here is going down to create the tweet as well. So that's the LinkedIn automation. While that's running in the background, I'm going to briefly show you the Twitter one. Now this one's really, really cool. So act as an informed Twitter expert in the industry. You will be provided with an article. You will create an optimized Twitter post based on the article, incorporating posts and best practices using the structure outlined below. Read the following article and create an engaging tweet based on its main points using the best tweet practices below. Concise, clear message. Use the format hoc, explain, question, hashtags. The hashtags help us reach a broader audience. The question will drive engagement for us. The hoc will get them to read it and explain or provide the value. We want to limit it to 280 characters maximum. Keep it factual, thought-provoking questions. Use a couple of hashtags. Never start sending with emergy, et cetera, et cetera. And then what we've done with this tweet bot is really cool. We've got example text. So this is an example of the text that you'd be provided. And this is an example of the kind of tweet that we want. Here's another example. And what the tweet would look like in that case. And, you know, we provide it with these specific examples so it knows exactly what we're looking for. And as you can see, it's gone through the whole process again, just like identified it, the Plexi got it. And if we come over to our LinkedIn bot here, for example, and we can scroll down, we can see here. Let me move across to the right so you can see a little bit better. Come down to content, plus one, text, value. And there we go. Big shift in the financial world. It explains what it does. It explains what it actually means. And then give some commentary. And then it gives some engaging questions. And it finishes with some hashtags. And look at this. Let's just discuss below. What are your thoughts on how major financial markets can have a significant change? That is phenomenal. I'm really happy with that. And that's our LinkedIn bot. And then we come back over to our tweet bot. We need to correct change the industry. So we're going to say in finance industry. By the way, guys, you would be surprised how like changing little features in the instructions tab actually makes a difference in terms of the output. It is honestly, honestly nuts. Cool. Superb. That's done. We're very, very happy with that. And then we're going to run this again for the exact same thing now with the Twitter post. And we can add in anything. We can add Instagram. We can add in Facebook. We can add any platform that you want. We've got the LinkedIn of a Twitter full guide and bot training. The Facebook and Instagram one will be coming very shortly as well. They'll all be available in the exact same place. And what you'll notice when you run these automations, by the way, is that if it forks off, it will go up and finish the entire route. Then it will come down and do the second route. So don't worry that, you know, this one doesn't look active yet. It has to wait until the first automation is finished. And the amount of time this is going to save you is ridiculous. Like think about the amount of time it takes to write up your own particular thoughts on something versus just running an automation that does this cross platforms. So this one's done. Let's see what the tweet looks like. And look at this guys. The exact same prompt. Exact same prompt. Is Shell bidding farewell to the London Stock Exchange? Shell plans to merge its entities, potentially moving its market cap from LSE to Amsterdam, thus possibly undermining London as a top global financial hub. Could this spell trouble for the UK financial market? Hashtag finance, hashtag stock market. That's a tweet that genuinely speaking is, you know, that you would engage with. Tweet is short, simple and effective. Next thing I'm going to do is create an image specifically to that, right? So to do that, we click on add new module. Let me just zoom out of here for you because it is a little bit all over the place. Great. And we come down to, we want Dali three. Cool. We click on this. We show more. And we're going to click on generate an image. Perfect. Using Dali. The model we're going to use is Dali three. That's the most powerful one that's available. And the prompt is going to be simply this. Generate an image based on this summary. And then we hit the thing here and then we're going to scroll down and find the perplexity output. Come down to usage. Come down to choices, excuse me. Come down to message and then content. So we come here. We're just going to duplicate this thing. So clone, come down, copy that over. And the thing we add to the prompt before we actually run the whole automation is the pixel sizes. So create an image based on this summary. Excellent. And you see, so for LinkedIn, it's 1200 by 627 is the best practice. Superb. And then we are going to copy that. Click okay. And then we come back down here. We're going to paste that. And then for Twitter, we want 1600 by 900. Excellent. And then we add in LinkedIn or X. And then before we continue with that, let's just run it one more time so we can see how good the images are that it's generated. Now you have two options when you complete this. One is you can just set up the text to publish directly on LinkedIn, X, Facebook, Instagram, if you want to. But if you want to look at it and say, I just want to get a list of 10, 12, you know, however many ideas they're all fully fleshed out. And I just want to edit it a little bit and then post it myself. All we do instead is we just change this LinkedIn, this X, this Facebook, this Instagram to a Google document. And we just add it to that Google document and all the tweets and LinkedIn posts will just build up on that page. And then you can just copy, paste it, download it and do all that yourself. So you can make any edits that you want to. I would recommend doing that initially just so you get comfortable with the inputs and the outputs. And when it's really nailing it for you, then just move it over to completely automatic. So we come here, for example, we come down to data array plus one. There's the prompt. There's the URL. We copy all the URL. And there we go. Is Shell trying to kill the London stock market? And there's the full photo that you want. And by the way, guys, you can basically change that however you want to. If you want to have it in a certain style or not, we can add all those things as well. I'll be releasing a kind of image generation best practice as well. But just for this video, I just wanted to show you the basic bones of it. I will add that to the LinkedIn doc and to the Twitter doc and the Facebook and Insta one that comes live. So you can optimize it and choose from all the different styles and artworks that you'd like with examples. So if you want to do it to a document first, here's exactly what you do. We delete the LinkedIn thing. We're going to add a router here. Perfect. Here we're going to come and we're going to look for Google Doc. Wonderful. And then we're going to click on insert paragraph to a document. You come down and choose your file. I've chosen content ideas file, which is absolutely fine. We click okay. Then we go to append a text. So what text do we want? All right, cool. So what we do is we're looking for this one here. We hover over it and you'll notice it flashing in the background when it's something to do with that image. So what we come down here is we go down to content, text, value. You see, that's the one that we want, value. Perfect. So what we've basically said here is, hey, I want you to put what that output is in this Google Doc. We click okay. Cool. Then we go to Google Docs again. Except for when we scroll down this time, insert image to a document. Brilliant. Well, Jack, what is the image? Well, first of all, let's choose the same Google Doc that we had to click here to choose a file. And then you'll see this image URL. So all we do for the image URL is we're looking for this guy here. So let's have a look at it. This whisper dolly. Again, you can see that it's pulsing, which is really cool. Come down to data and it's this URL here. Perfect. Excellent. Cool. Now I'm going to hit run once and it's going to run through the whole thing. And it's just phenomenal, isn't it? Just how much we can automate with this. Just such a good quality. And if you're somebody who's really particular on your writing style, you'll get very close with this problem that we've got. And then you can do the tweaks within a Google Doc. I mean, it literally be two minute exercise. Jump into Google Doc. That sounds great. That sounds great. Post it or maybe share it onto anybody that you like. But this can be completely fully automated. As you can see, it says it's complete and it says it's complete. So the moment of truth is check the Google Doc. Here we go. Wow. And now we have it. Guys, this is the full thing. Look at this. Big moves versus big blue. Is Shell's potential exit a wake up call for the London stock market? As one of the giants in the energy sector, Shell's recent announcement regarding its plans to block. For a LinkedIn post, guys, this is phenomenal. That looks like it spent a lot of time generating it. But let's check out the image. I'm curious what this looks like. Wow. That is epic, right? Like that is absolutely insane. Because you can see the Shell thing, right? Like it's got this stuff. It's to do with the stock market. So you've got the Shell coming out of it. There's helicopters and stuff. That's all automated. All we've done is give one single prompt to Slack and sent it through the entire journey. It's created that. And it'll do the exact same thing for Twitter, Facebook, Instagram, you name it. And creating social media content is amazing. But if you haven't made your own AI brain yet, I'd really encourage you to watch this video here because it goes into a lot of detail about it. And tell me down below as well what you'd love to see automations on. I've got some pretty incredible ideas about how you can create a full process for getting a six-figure roll in terms of scraping the data or making the social media scraping automatic so you don't need to include the URL. Let me know your thoughts down below. In any case, guys, have a wonderful week and I'll see you next time.