My 8 Most Used AI Tools For Content Creation

Today, I'm going to be answering the number one question I keep getting, and that is, Igor, out of all of these AI tools, which ones do you actually use most of the time? And within them, how do you use them? Well, to be honest, the answer is quite simple. Two thirds of my time I spent with ChatGPT. And that's why I'll be dedicating this video to my ChatGPT use cases for content creation. And concretely, I'll be giving you some of the quickest wins and ways to get advanced results very fast. What I won't be covering is my second big use case, which is image and video generation. This is going to be all about text generation with ChatGPT and some other large language models. So with that being said, let's dive into some actual use cases that I wouldn't want to live without anymore. And look, let's just start with a big obvious one. ChatGPT. Obviously I use this all the time, but how do I use this? I think that's the interesting question here. Well, a few things come to mind. First of all, idea generation, and there's about a hundred approaches to this. Okay. But I usually return to two. First one is the technique that you might be familiar with. It's called multi shot prompting. And that is when you start with something simple like give me ideas for YouTube videos, but then you actually provide it with free or more examples. So, for example, I'll do my three most popular YouTube videos and then it's going to generate a whole new set of ideas. Okay? And this is really my secret sauce to generating ideas. You got to infuse it with your custom data, with your information. So there you go, a bit more of a custom approach. If you have even more data, you can always go to the advanced data analytics, which was previously called code interpreter. Now, you could go ahead and upload any type of data here, for example, hundreds of YouTube videos, and ask for further ideas just like the ones you provided it with. Okay, and here's a bonus one where I initially thought that this goes beyond the scope of this video, but actually I keep using this more and more. And it's a script that I run on my laptop and just bear with me for a minute. You don't need any coding skills to understand this. And what it does is it deploys multiple ChatGPT agents. So it's just multiple copies of ChatGPT talking to each other, each one with a distinct personality. But here's the kicker. I started using this one more and more ever since they introduced function calling, what that means is that these different personalities get to do things like search Google or my favorite one, search YouTube. You can actually quite easily hook up the YouTube API API into the script. And all of a sudden one of the agents is just sitting there searching YouTube and providing the other agents with information. Now, by no stretch of the imagination do I use this daily, but I find myself returning to it like once a week and just running it. And then from time to time I come back and change some of the personalities and look for different results. And I just find this to be a fantastic way of brainstorming. That works for me because I get to follow the conversation they're having and their train of thought and their entire discussion. At first I thought of it as a gimmick, but it's actually quite practical and these things are only getting better. And my favorite part about this is that we made this work inside of our team chat in the AI Advantage Discord. So all team members now get to go in and they get to call these agents and they get to task them with different missions without knowing how to code. So all they need to do is type in a simple command and tell them the goal and then the agents go out and they complete it. Look, we're not at a point date where this would be a complete game changer, but I'm very happy with how it keeps the entire team on top of all things AI. This is one of the bleeding edge possibilities, but that not just me, but 10 people experiment with regularly. I'll definitely report back if we come up with something acutely practical. All right, and here's another GPT4 powered way how to enhance your content on LinkedIn Caplio. What it does is use GPT4 to maximize your reach, engagement and followers on LinkedIn all in just 10 minutes a day. And the possibilities are truly vast here because it doesn't just analyze your content, it also understands your audience behavior and creates content for you on autopilot. But don't just take my word for it. Over 6,200 LinkedIn experts are using it daily to enhance their content game on LinkedIn. And it should be said that this works equally as well for both professionals and brands. So not just that it creates the content, but it also schedules them and you can dive into the analytics of it all, which is essential. Oh, and on top of that they have a chrome extension called Taplio X which seamlessly integrates into your LinkedIn experience, making it even easier to use. And what I personally really like is that it doesn't just help you reach new people, it also helps you manage the leads and contact them all inside of one app. So if you want to both find and engage with the right people to turn followers into business opportunities, Taplio's got you covered. So go ahead and start leveraging AI for your LinkedIn growth today. And guess what? You can start completely for free. Just check out the link in the description below. And now let's get back to the video. So the next use case I actually return to regularly is rewriting text or generating drafts. And the reason I love this so much is because I don't use these AI tools to write for me. I use them to assist me in writing. Okay, so usually I let it generate two to three drafts and different voices and tones. I just over time develop certain presets that I like and then I use those as inspiration for my very own writing. And then once I come up with my own writing, I like to rewrite. With ChatGPT, this just makes a lot of sense because you're not giving it a blank slate and telling it, hey, come up with something from scratch. No, you're giving it a structure, you're giving it the content, you gave it a lot of directions, and then it just does the fine tuning for you. And as you can see, this is a common theme here. A lot of the best ChatGPT use cases are hidden behind providing it some of your own information, not just going straight from a blank canvas like this to a result. No, bring in examples, bring in data, give it your own writing, and then prompt on top of that. That's the real power of it. So, for example, let's just take this video description and let's just say rewrite this video description to be more engaging and more concise. And then I just paste it in and there you go, it's going to do exactly that. And this is just so useful because whatever you do with content, you're always going to be faced with writing. And like this, you can just keep iterating over a certain idea until you're happy with it without using too much creativity, which can be exhausting at times and some people are just not that good at it. So there you go. So here's a practical example that I used for one of my shorts. And as you can see, super simple. I just rephrase, then specify the content format. And I like to use emotionally engaging. Plus I get some alternatives. And there you go. Instantly have free alternatives to choose from. So this is very powerful, especially if you're working by yourself, because like this, you're going to get another opinion that you can now weigh and contrast with your very own ideas and thoughts. Really, if you think about it, the $20 a month for this is a steal. But you probably know that already. All right, and here's a little bonus tip from editing Igor, because by now everybody has access to ChatGPT vision. And I need to point out that a lot of the things I'm showing here are all about simple prompts with some context that you're giving it. So either an Excel sheet with all the titles, or you give it multiple examples to reproduce those examples, or we're using custom instructions to give it all our personal context so it can make better recommendations. But actually one of the best ways to give it context is visual. And especially for content creation, use cases. You could do things like this. You could go to any page and you can screenshot it. And now you go into ChatGPT, you give it that image and you ask, what could I improve about my newsletter signup landing page? We'll just ignore that typo. And there it is. It just gives you a full analysis. It can read the content, it understands what it does. And because it has all the info on me and what I do pre saved, it has that context, it understands what I do, what my mission is. So as per usual, the more context you can give it, the better results you will get. One quick example, which has been going a bit viral on Twitter and LinkedIn the past few days, and this was actually one of the first things I tested once I got access to it. So you can just go to social media profiles, screenshot them, and ask for recommendations on further ideas. And this works not just on YouTube. I just attach the screenshot and I'll just say, give me more video ideas. And again, it pulls in the context from the custom instructions. It uses this picture and all the rich information. So the view counts the thumbnails, the titles, and there you go. These ideas are just at a different quality and precision level than anything you would get from a simple prompt, because it's getting the context from here, from here. And then all we need is a simple prompt because it already has all the details it needs. So there you go, vision for content creation. Absolutely massive. Okay, and to round this section out, I want to go beyond ChatGPT, because sometimes I do find myself using other large language models, as you might have seen by the videos I upload on the channel recently. And here's the thing, there's More and more of them out there, but none of them beat GPT4 on all categories. That's why it's still a default. But there are certain strengths you want to consider, and in certain situations you want to lean into some of these tools. Because something like Claude 2, which by the way is still not available and you need to use a VPN to access it, can take large context windows, so you can upload research papers and talk to those and use any prompts that you want on top of those documents. So CLAUDE is really nice for the context window. And then new open source models like I talked about on a channel like Falcon or Llama, they just have a bit of a different flavor. So sometimes if I want to switch things up, I just take some of these basic prompts in here and I open up a second window with something like the Falcon 180B demo on hugging Face here, where I can just use it for free, right? And I just submit the same prompt and there you go. Now you can contrast the results. There you go. Those are my most frequented use cases for large language models that generate text. Nothing crazy, nothing super exotic, just the basics done in a way that are customized to my own needs. And with that being said, I need to point out one more thing, and that is that I use custom instructions like said I all the time. And that's what in practice makes some of these simple prompts so powerful. Because I have these custom custom instructions that communicate everything about my channel that I needed to know. The AI advantage now offers that. There's a link in the description to check that out. Okay, here's another one that is a staple of my workflow at this point. For every video just like this that gets created, I have a workflow in place that takes the video, turns it into a transcript, and then on top of that transcript runs tens of prompts and saves all that. And then various team members that need these different things have access to it. For example, just in the YouTube folder, I have an auto description that gets generated for every single video. And honestly, this is just copy paste ready here. I don't use it for drafting. These just work and they're SEO optimized. Another one is this title and thumbnail generator. So on every video that I create, although the title and thumbnail might be ready already, here, it just spits out 10 more ideas, making my thumbnail designer's life way easier if I ask for some alternatives. And here's a fun little bonus tip that I never mentioned anywhere before. But what if you just took one of the transcripts of your videos or you speaking or you writing and then you just asked give me actionable suggestions for how I can improve my YouTube videos based on this transcript. And yet again as you can see we're using a mix of custom data and a quite simple prompt and look at that bunch of suggestions concretely tailored towards me. And as you can see, these customized outputs are where the real value here lies, right? You just include all the relevant details to you and your needs and then every idea it generates, every draft it creates, every text that it rewrites and any other task that you might be tackling within a large language model will become customized towards you. So keep that in mind whenever you use GPT 3.5 or GPT 4. And if you're looking for even more content creation use cases but more on the visual side AKA images and video, well this video does exactly that. I'll see you there.