A (0:00)

What I'm about to tell you might be a surprise today. Right Now, I use AI to automate more than 50% of what I do. And this is coming from someone who's generated millions of dollars with my writing. I've published 10 books. I've co founded some of the biggest writing programs on the Internet. I've figured out how to write such powerful AI prompts for ChatGPT and Claude and Gemini that I don't need to do all the writing anymore, which is what I want to share with you today. I use it to write social content, writing weekly newsletters to our different email lists. I use it to write ad scripts, video scripts. I use it for fiction writing. I've probably spent over 2000 hours at this point experimenting with these different AI platforms, writing and rewriting my prompts. And let me tell you, if you aren't getting the outputs that you want, it's not because AI isn't capable, it's because you aren't prompting it correctly. So in this video, I'm going to give you nine ChatGpt and Claude writing tips to get crazy good outputs. Once I started doing these things, the outputs that I was getting completely changed. So let's dive in. So after playing with AI for two years now, I've really realized that when it comes to prompt writing, there's three different ways to interact or prompt AI. There's rabbit hole prompting, which I'll explain in a second, there's brainstorming prompting, and then there's automation prompting. So let me give you some examples of each one, because all three of these work, they just work in different ways and there's different benefits for each one. So rabbit hole prompting is when you just want to learn about something. It usually comes down to you wanting to educate yourself. So for example, I might say, tell me about the Roman Empire, right? So this is sort of the equivalent that the modern day equivalent of going down Wikipedia rabbit holes. Right. And so the way I like thinking about this is it's very conversational. It's you start with some question or some topic, you say, tell me about this thing or I want to learn, you know, why does this happen? And then based on the output that it gives you back, you pick one of those things to go further on. Right? Right. So here if I take a look at our example, maybe we go military power. Oh, that's really interesting. So I might say, interesting. Can you tell me more about what made the Roman Legion's military powers so powerful and unique? So what's happening right is you're picking a topic and then it gives you a bunch of information and then you pick a smaller piece of what it just gave you and then it gives you a bunch more information and then you pick a smaller piece within that and you just keep going down the rabbit hole. Why is this valuable? Well, I mean this is really just an advanced form of Google, right? It's, it's what we were doing in search engines for the past 20 years. This is just faster and more linear. You're just going from point to point. You're just going further down the rabbit hole. Right? This is actually how most people use AI and there's nothing wrong with this there. I use this every single day, especially on my phone. If you don't have the ChatGPT app. I love the voice function. So I'm always asking questions, um, especially when I, I think of something and I'm like, hey, real quick, tell me about this or what should I do about this or how should I make this decision? Gives me some information. Oh, I have a follow up question. That's rabbit hole prompting. There's nothing wrong with that. There, there are a lot of use cases where that makes sense. But this is not how you get the highest quality writing or the highest quality outputs from AI to incorporate into your own work. You know, especially if you're using AI for writing services, you're using it to work with clients, or even if you're just using it to write for yourself. You want help writing your own social, help writing your own newsletters, you're not going to get the output that you're looking for by just asking like one or two sentence questions or giving it these really short prompts. Right? So that's the first way to use AI is rabbit hole prompting. I would use that primarily for just self education. The second way that you can use AI is what I like to call brainstorming prompting, which is where you're actively trying to think through something and you want a thinking partner. Now this is a little different than rabbit hole prompting because rabbit hole prompting is more, you're trying to learn more about a pre exist and so you're really just using AI, sort of like a, like a modern day technological library. Brainstorming is different. Brainstorming is where you have to go a little bit more in depth and you're trying to articulate your ideas and you basically want to use AI to stress test those ideas. So for example, I use brainstorming prompting all the time, both on the nonfiction side when I'm Trying to outline a book or when I'm trying to outline a new digital product. Or maybe I'm trying to figure out how to reassemble a module inside one of our curriculums. Right. I'm actively thinking through how I want to create something or change something or improve something. I also use this on the fiction side. So I use this all the time for stress testing story ideas or looking for holes in a plot or brainstorming world building. What are different attributes that I could create? Let me give you some examples of how this works. So let's say we're trying to create a new digital product. So let's say it's May every November. We want to create some sort of product for a Black Friday drop. So usually, you know, six months out I'll start brainstorming what are some different digital product ideas? Well, I might say I'd like your help brainstorming some digital product ideas and then I want to give it a little context. A digital product is a text and video course priced at $350. In the past I have always created writing related digital products like category newsletter creator which helps people start and scale their first free newsletter or low ticket launch pad which helps people start and launch their first paid digital product. I have covered so many writing topics at this point I could use some help generating net new digital product ideas. My name is Nicholas Cole because AI can look people up on the Internet, right? So if I give it this context it's going to immediately know oh well I have all this other context on you. My name is Nicholas Cole and I am also the co founder of ship 3430 and premium ghostwriting academy. Can you please give me 20 net new digital product ideas that are still in the world of writing, digital writing, ghost writing, but are unique things I haven't created before and or new niches we could expand into. So if you notice even this prompt is still like relatively in depth. I'm being very detailed and as we go along here, I'm going to explain some of the decisions that I make whenever I'm prompting. But even with brainstorming, you always want to give as much context as possible. You know you're not going to get the highest quality outputs when you're just giving a two sentence prompt like help me come up with ideas for an essay. Help me come up with ideas for a digital product. AI doesn't have a direction to run in but if you give it really specific instructions and not just instructions, but context, AI is going to generate Higher quality ideas. So let's see what it comes up with. Technical storytelling Academy. Maybe the financial writers blueprint. That's niche down like for financial writers. I don't know that I would want to do that. Voice to page mastery. That could be kind of cool. Is walking writers through instead of writing but transcribing through voice. That could be cool. Digital PR playbook. I don't love pr. The memoir Ghostwriter. That could be cool. I mean it, it's a very different market, but it's an option. Script writing for social. We haven't done that yet. Actually that's a, that's a good idea. Would be creating a low ticket digital product around how to write scripts. AI writing collaboration. Yeah, I mean anything paired with AI and writing these days is performing well. So that's always on the table. Podcast scripting and maybe sales page psychology. That could be really good. We have some resources on that. But leaning into that would be really good. Right. So notice like there are a bunch of really good ideas here, but the only reason that it's giving me some of these really specific ideas is because I prompted it with this level of context. You could do the same thing on the fiction side. So for example, maybe I say I am working on a new fiction series and it is a. Let's say an ancient civilization is about an ancient civilization that discovers world changing technology that has been buried beneath the ground for centuries and now its power has been discovered and society needs to figure out what to do with it. I'm just riffing this. That's actually a pretty cool idea. Let's see. I would like your help brainstorming the world. Let's do some world building. Now this prompt, because it's so short and again I'm going to explain this as we go along because these are really important nuances. What's happening is, yes, I'm giving it direction, but I'm actually asking AI to fill in a lot of the gaps. I'm asking AI to lean on its own intuition. And what I mean by that is I'm saying help me brainstorm the world. Let's do some worldbuilding. Which means AI has to sort of ask itself. Okay, what do I know about world building? You will get a decent output with this. The way you get a 10x output is by saying, and when we do world building, I would like you to fill in these specific attributes. So let me show you what that looks like, you know. But when world building, I would like you to specifically generate ideas in the following categories. So I want to know, you know, on our fictional planet, what are the unique attributes of the religion of, let's see, the music culture of rare artifacts and collectibles, let's say borderline mystical or magical powers through technology, major cities, methods of transportation, food and cuisine, social hierarchy, government structure. This is going to get us a way higher quality output. And the reason is because we are being very specific about the attributes of world building that we would like help brainstorming. So again, one of the things that you constantly want to be asking yourself when prompting is you want to ask, how can I be more specific and how can I be more objective and actually remove as much of AI needing to lean on its own intuition? Because other like that's why low quality outputs happen, is you're asking AI to fill in too many of the gaps, you actually want to remove as many of those as possible. So let's see what it comes up with. Awesome. So you can see it's starting to fill in. It's giving a bunch of sub variables inside each of the categories that I listed out. So inside religion, inside music culture, inside rare artifacts, collectibles, et cetera. But this sort of leads to the next question, which is, okay, but how would we get an even higher quality output? Right. Because when you look at these different categories, yeah, I mean these categories and subcategories make sense high level. But there's nothing actually super specific about the story here. This is where brainstorming prompting compounds on itself. So how you would get more specific or higher quality outputs is here in the prompt. You wouldn't just say, oh, I'm working on a fiction series about an ancient civilization that discovers game changing technology. That's very high level and that's very vague. The way that you would really dig into brainstorming is you would dump in. Here's a three page synopsis of all the nuances of my story. The arc, the main character, where I see this going, some of the unique attributes that I've come up with. The more context that you give AI, the easier it will be for them to help brainstorm ideas. And this is where brainstorming, prompting and rabbit hole prompting can sort of work side by side. Brainstorming benefits more from context. The more context and information you can give it, the higher quality ideas it will spit back to you. But then as it gives you ideas, you can go down the rabbit hole, you can interact with it. So for example, you might say, this is what my story's about. These are the different things I want your help World building. And then it gives you a bunch of ideas and you go, oh, that one piece that you gave me is really interesting. What are some other options of that? Give me 10 more options or 10 more examples of that specific thing that you just gave me an idea for. So that's where these two can go hand in hand. Okay. The third is automation prompting. And this is where things get the most advanced. This is where you have the most leverage. Okay. Because rabbit hole prompting and brainstorming prompting are more interactive. It's more. You discover it as you go. Automation prompting is usually what happens at the end of rabbit hole prompting and brainstorming prompting. Okay. Automation prompting is when you take something that you've already learned how to do manually and you reverse engineer it into an automation. And in this case, an automation is just a prompt that you could copy, paste to replicate a result over and over and over again. So this is where prompts are really a lot of upfront work that give you leverage on a task indefinitely. That's why they're so valuable. Everyone thinks that the bottleneck to writing really great prompts is knowing how to use AI. And that's not actually the bottleneck. The bottleneck is knowing how to do something manually with so much clarity that you are able to articulate how you want AI to do it for you. And this goes, you know, this speaks to a bigger idea that I've been thinking about a lot, which is everyone wants to automate things they've never done before. And really what I have been saying back to writers is you can't automate what you can't articulate. So automation prompting is the result of learning how to do something manually and having a lot of clarity over how it works. So let me show you an example. Here are a bunch of different prompts that I've created inside of our premium ghostwriting academy. So we give these to writers inside PGA because we want to help them not just understand and learn the skills manually, but we also then want to say, and here's how you get leverage on those skills. And a lot of times I am sitting down and taking 10, 15 years of knowledge, and I'm compressing that into a single prompt that then all of our students can copy, paste, and use for themselves. Right? It's pretty cool. So let me show you how this works. I built this prompt to write short form social content. Now, why did I write this? Well, because we are also an AI first company, and so we are constantly asking ourselves, how can we get more leverage on our skills and you know, when me and my co founder, Dickie, and when our team, you know, we have a staff of writers and marketers internally as well. We have a team of 30 plus people at this point, when we sit down to engineer social content every week, we're really just going through the same formats in our head manually. If I was to train you on how to write really effective short form social content, like content on X, formerly known as Twitter, for example, I what I would do is I wouldn't just train you on good writing, I would train you on formats that perform well, and then I would train you on the attributes and the rules of each format, and then I would give you examples of each format so that you could build your own pattern recognition. Right. I would be as tangible and specific as possible if I was training you, another human being, to do and create similar types of content. AI is no different. The same way that you would articulate it to a student or an employee or a team member or an intern is the same way that you would articulate it to the technology. And so a lot of the skill isn't knowing how to use AI. I mean, AI platforms are really easy. It's a text box, right. The skill is knowing how to do something manually with clarity and being able to articulate how it works so that you can get leverage on that skill. So let me show you what this looks like. This is a mega prompt that I created to write short form social content. This mega prompt is essentially training AI on all of the same principles that I train our internal writers on and that I train all the writers in our premium ghost trading academy on as well. Right. So how this prompt works, just to show you is I'm giving some very tangible instructions here at the beginning. You're a social content generator. I want you to generate short form social content that fits within 280 characters or less. And I'd like you to structure it in the format of a day. So here are the different time slots and I would like you to write one short form piece of content in each day. Where most people stop is at the end of this little instruction box. They're like, hey, just write really great social content for me. But that is like the tip of the iceberg. You can push this so much further by saying, not only do I want you to write short form social content, and not only do I want it in these time slots, but I want it in these specific formats and in each format. Here are the rules of each format and here are exact examples of each format. This is where, you know, it probably took me three or four hours to write this entire prompt. Now, you know, why did it take me only four hours? Well, because I've also been writing short form social content, and I've been writing about how to write short form social content for a decade. So it's actually not four hours, it's, you know, hundreds, and if not thousands of hours. And then I just sat down and compressed it all into a prompt in four hours in one day. Right. But all of that upfront work now means that every time I or a team member sits down to write short form social content, it will generate a similar result over and over and over and over and over again. That is a tremendous amount of leverage. This is what automation prompting is. It's where you spend a lot of time up front creating a prompt that you can just copy, paste, and use over and over and over again. So I'm using Claude here. You could use ChatGPT, you could use Gemini. They all work similarly. I copy paste it into Claude, it says, great, I'm ready to help you generate a day's worth of thought leadership short form social content. All right, so let's say my topic is getting started as a ghostwriter for business owners. This is typically what I would write short form social content on. So let's see what it comes back with. So this is pretty cool. So these are actually different formats that I'm training it on. I train it on a form, one format called paragraph style. I train it on another format called what? How? Why? So each of these formats inside the prompt have their own rules, their own attributes, their own examples. And so it's not just generating content based on AI's random intuition, it's generating it based on all the specific rules that I have trained it on. And if you look at a post like this, if I would post this on X, you wouldn't know that it was written by AI. It's great. I mean, maybe I might change one word or two words, but, you know, 95%, that's exactly the sort of piece that I would create. And so you have to think like, this is the real benefit of automation prompting. This is where things really start getting out of control. Because if you take the time, you take 5 hours or 10 hours to assemble a prompt, well, now you just cut the amount of time that it takes you to do that task over and over and over again by 50% or 80% indefinitely. So these are the three ways to use and prompt AI. And every time you sit down to write and write with AI, you want to first ask yourself what sort of prompting am I doing here? Am I using rabbit hole prompting? Am I just trying to learn about a topic? Am I doing brainstorming, prompting where I want to give appropriate context, I want AI to give me ideas back and then we're going to push on those ideas and expand those ideas together, or am I doing automation prompting? And automation prompting is usually the end result of all the work that you do up until that point. By the way, if learning how to write with AI is a priority for you, I would really encourage you to check out our newsletter on Substack. Right? With AI, it's actually the number two education newsletter on Substack. And every single week we basically share a different writing framework and we share a prompt that goes with it. So not only do we explain how to do something, right, the manual version, but we also then give you the prompt for how we have figured out how to automate that as well. So check out Write with AI. I think it'll be a great resource for you. AI writing Tip number two is I highly recommend you use Claude Projects. So the big three AI platforms are Claude, ChatGPT and Gemini. I always get asked the question, which one is better? The honest reality is that these platforms are changing and their models are changing so quickly and all the time that they're. Whatever answer I give you today might not be correct tomorrow. So I would just encourage you to play with all three. Very often I will take a prompt and just copy paste it into all three and then I just look at the different outputs that I'm getting. Experiment. CLAUDE has this cool feature though, called Projects, and it's here on the left hand side when you're signed in. And think of a project as a bucket. And inside that bucket is all the relevant information for that specific project. Right. So why is this helpful? Well, oftentimes the way that you get higher quality outputs when writing with AI comes down to the amount of context that you give it. So more in depth prompt equals better output. More in depth prompt plus supplementary assets or education or resources equals even higher quality output. Right? So the more context that you give the model, the more likely it is to give you the thing that you're looking for. So if you think of a project as just context, what's really happening is you're saying, hey, AI, I want you to use all the powers that you have access to. However, as we work on this thing, I want you to keep the direction inside of this bucket. I don't want you pulling from every possible corner of the Internet. I want you thinking in the direction of all the context that I've put inside this project. So for example, whenever I'm working on a new book, I might start a project for that book. If I'm working on fiction and I'm working on world building, I will create a project for that story or for that world. If I'm working on a new product, I will create a project for that product. And the reason is because I want to keep dumping context and information inside of that project. So for example, let's say I'm working on a new low ticket digital product and it's going to be organic short form video script writing digital product. And this will be a 10 to 15 module text and video course teaching people how to script organic viral worthy short form videos. Three minutes or less create project. Okay, great. So here on the left, this is our typical chat box. So no different than if you go to ChatGPT or Claude and you open a new chat, right? It's a new window, it's a new context window and you can interact with AI that way. When you're interacting though, in this chat box inside of a Claude project, again, you're keeping the conversation within the bounds of that project, right? So it just helps AI laser in on this is, this is exactly what we're working on. I'm not going to get distracted by every other possibility out there. Here on the right you have project knowledge and so what you can do is you can dump in PDFs, you can dump in Word documents, you can just add resources here that give more context to the project. Now you can do something similar inside ChatGPT or a basic CLAUDE window. And, and what you can do is you can just start typing and then drop in documents and you can sort of do the same thing. But I have found that CLAUDE projects this really helps organize it. And for some reason the, the outputs are way higher quality because I think these projects are engineered to keep all the thinking within a container. And one more pro tip here, this is really cool as you go along, whatever the project is, if you're writing a book, if you're building a digital product, if you're working with a client, if you're writing a fiction story, as you go along with the project, each decision that you make or each thing that you finish, you finish a chapter, you finish a module, you finish creating a world building document, you can add it back into the project here. So I'll Give you a quick cool example. I'm working on a fiction series right now and I used a lot of brainstorming, prompting to come up with ideas for this story. Then on my own, I created my own manual world building document. Then I did more brainstorming with AI and I said, hey, give me more ideas, help me flesh this out, right? And ultimately what I ended up creating was not just a world building document, but I created a map of the world. Then I created sub documents summarizing the major cities in the world. Then I created sub documents of here are my different characters and here are the attributes that make each character. Right? A lot of upfront work. As I crystallized each piece, I dumped it back into the project knowledge here. And so then as I interact with the CLAUDE project and as I write chapter one, chapter two, chapter three, the model has all the context it knows. Oh, you're referencing this major city. I, I can see that major city because it's in one of those documents you gave me. And not only do I see the major city, but I understand all the attributes that make that city that city because it's all in the context window. And so that's where then like storytelling and article writing and all of these things become way more powerful because you're not just asking AI to come up with some answer on the fly. AI has the context of, I understand this world, I understand the rules that make each city its own thing. And so that's why as you build a product, or as you write a book, or as you assemble a story or whatever it is that you're doing, you want to keep adding more information back into the CLAUDE project so that it has continued context. The same is true when you get in the writing. When you finish chapter one, you want to give chapter one to the CLAUDE project before you start writing chapter two, because then it can read it and go, oh, all these things just happened in chapter one. Now I have more context going into chapter two. This is where things start to compound and get really cool AI writing Tip number three is you have to understand the difference between singular prompts and modular prompts. Okay, so little crash course on AI and how to get the most out of AI, us everyday people. We are using just the front end of ChatGPT and Claude and Gemini. This and the front end is very user friendly. It's just a text box. The reality though is these platforms also have backend systems. They have, they have backend infrastructure. This is where having knowledge over code is really powerful because chatbots, tools on the back end are way more powerful than just the front end. Now, for 99% of use cases, especially when it comes to writing, you don't actually need to know how to code. You don't need to tap into any of those things. You can, but it's just not necessary. You can get amazing outputs just by using the UI on the front end. However, I've spent a lot of time trying to understand the differences between the front end and the back end because I want to know how to get the most out of AI. And one of the big ideas that AI has educated me on is that when it comes to prompting, there's sort of two schools of thought or there's two approaches. There's, you know, singular prompts and modular prompts. A singular prompt is typically, you want to think of it almost like an atomic essay or a, or a short form essay, probably looking at something in the ballpark of like 500 to 1000, maybe 1500 words. The goal of a singular prompt is to accomplish a single task. It's not just that. The longer the prompt, the better the output. That is not what I'm saying. Something I tell writers all the time is, you know, word count is a horrible measure of value. Where prompts go wrong is they try to do too many things at once. So an easy example would be, imagine if I tried to create a prompt that was something like, I'm going to, or I'm asking for your help brainstorming digital product ideas. And then when we have an idea, I want you to turn it into an outline. And then from that outline, I want you to outline each individual module. And then after you outline each module, I want you to write each module. Those are four completely different pieces. And so it actually doesn't make sense to fit all of those pieces into one prompt, you know, especially because the goal, the way that you would get the highest quality output is you would want to go back and forth with AI. You don't want to just leave it to its own devices and be like, you know, a lot of times people ask, you know, how do I get AI to write an entire bookmark for me? Well, the way that you get AI to write a whole book for you is by breaking the book apart into a hundred substeps. You're not going to get a very high quality book or end result if you just say, go write a book, because there's too many sequential decisions in there, right? So when you're thinking about writing prompts, you actually don't want to Ask AI to do lots of different things all at the same time. Singular prompts are again, think of them like atomic essays or short essays. You want them to have only the context required to accomplish a singular task. Anything else is unnecessary and it actually leads to a worse output because AI has too many things to focus on. So most of the time, especially on the front end, you're going to be writing singular prompts. So anytime you sit down, you don't want to try and go, how do I cram as many things as possible into one prompt. You want to break something apart manually, first yourself, and then write smaller prompts for each individual task. The second way of prompting is called modular prompting. And this is where things can get a little more advanced. And I just want to plant the seed here. If you want to go fully down the rabbit hole, you know, I encourage you to explore this. Modular prompting is basically where on the back end, if you were to combine writing prompts with software, what you would do is you would basically extract as many variables or as many tasks as possible, and then you would write lots of tiny prompts so that AI knows when to pull in each prompt for each step. Okay, so I'll give you an example. Say that you wanted to build a AI writing platform to help people write nonfiction thought leadership books. Well, you wouldn't write some, you know, 30,000 word prompt that you just paste into the front end of ChatGPT and then you say, go write me a thought leadership nonfiction book. That's not what you would do. What you would do is manually you would break apart writing a book like that into as many smaller pieces as possible. So what would that look like? Well, first you would probably come up with a really compelling market to write a book inside. Maybe there's a bunch of research that goes into that. Then once you pick the industry or the market that you want to write within, you would come up with a main title and subtitle. Well, you don't just say, hey, come up with a main title and subtitle. You would want to give AI frameworks for assembling that. That's a separate task. Then when you have a main title and a subtitle, you would want to flush out an outline. Well, you don't just say, hey, go create an outline. There are rules for creating an outline. There are styles for creating outlines. So what styles or formats do you want to train AI on how to do? Then after you create the outline, then you want to draft the first chapter. You want to clarify how it's Structured. You want to clarify the topic, you want to clarify the voice, you want to clarify the writing style, you want to clarify the word choice. Each one of those is its own variable and subtask. So modular prompting is where manually, you separate all these different things and then you write singular prompts for each individual task. And the reason it's called modular prompting is because on the back end, with technology, you can then automate and say it's, you know, if, then if we're at this step, these are the different prompts that I want you to pull from. Once those are completed, then I want you to pull from these prompts. Then I want you to pull from these prompts. So it's not one giant task. It's actually a task broken down into smaller tasks, broken down into the sub variables that make each task successful. Automated on the backend. This is where things get more advanced. I just want to plant the seed for you here. This is the tip of the iceberg. There are so many cool things that you can do with AI and writing with AI. And the big takeaway here is that when you sit down to write, you do not want to try and compress, you know, eight different tasks into one prompt. You want to look at something that has eight different components and go, I need to write eight different prompts. And if you're just using the front end of ChatGPT or Claude, you would just run each prompt individually. AI writing tip number four is you want to name the different formats that you're using. You always want to be naming things. So a very simple example is let's say you're creating a prompt to write short form social content. Okay, great. We're going to call this, you know, you are a social content generator. Or let's say you want to write a prompt for writing newsletters. Or let's say you want to write a prompt to reach out to ghostwriting leads. Okay, you are an outreach specialist for ghostwriters. Or let's say you want to write a prompt for coming up with how to package and price your services. All right, you are a ghostwriting services packaging price pro. Why is naming so important? First of all, I think it's a great mental exercise as a writer when you're writing a prompt to be very clear, both within yourself, but also AI. This is what we're doing here. You know, when you give something a name, all of a sudden it has direction. So I do think that that's. It's just a helpful forcing function for you and for the technology. But second is as your prompts get a little bit more advanced, naming and naming formats makes it very easy for AI to follow along. So I'll give you an example inside my short form Social Content Generator, I train AI on five different formats. So these five formats are formats that I was already writing with. They were already things that I found myself doing over and over and over again. And now I can get leverage on that skill by just taking four hours and building a prompt and formalizing it. However, if I was to just write this prompt and say, I'm going to train you on these four different or five different formats, but I'm not going to name each format, it's actually going to be kind of hard for AI to follow along. Wait, which format are we talking about? Right. They're going to get lost in formatting soup. And so there's a lot of clarity that happens both, again, you as the writer, but also with the technology. When you say, I'm going to train you on five formats, and each format has its own name. And then as the prompt goes along, you reference the name and then you say in this prompt, paragraph style. Here are the attributes that make this format work. Well, here are examples you notice here. Here are a couple examples of format one, paragraph style. So you understand how this looks, how this is supposed to be. And so whenever you're giving things names, it makes it very easy for AI to follow along. And so this is just a best practice. Every time you're sitting down to write prompts, you always want to be asking yourself, wait, I'm explaining something. Anytime I'm explaining something, I should give that something a name. AI writing tip number five, you always want to use objective language. Okay, so first, what is the opposite? The opposite is subjective language. And what that means is I say something and that something is open to interpretation, which means If I ask 10 different people what that something means, I'm going to get 10 different answers. So an easy example would be, let's say I was writing a prompt, and in that prompt I said, hey, Claude, chatgpt, I would love your help writing really great social content that is very subjective language. The problem is that AI doesn't know how to interpret. Really great. Just like If I asked 10 different people, what, what do you think I mean by really great social content? I'm going to get 10 different answers. And so you want to avoid subjective language as much as possible, which means you want to use objective language as much as possible. Objective language requires no interpretation. And again, AI is technology. It's actually not thinking on its own. All it's doing is just scrolling through pattern recognition. So the more objective you can be with your instructions, the more likely it is that AI is going to give you what you're looking for. So let me give you an example. You are a social content generator. I would like you to generate a week's worth of thought leadership short form social content that will fit inside a standalone x post 280 characters or less. I'm going to give you the topic and I would like you to present a full day's worth of short form social content to me in the following order. In these five time slots, one piece of short form social content in each time slot. There is nothing subjective about these instructions. They are very objective. AI either writes short form social content 280 characters or less and presents it in these time slots, or it doesn't. It's very objective. But even this isn't specific enough, right? Because really what we want to do is we want to say, hey, these are the rules of what I'm looking for. But even this isn't enough, we can take it a step further. Not only are we saying, objectively help write short form social content in these timeslots, but we're also saying, and the short form content that you create needs to be in these five formats. But it's not enough to just say these five formats. We also want to say, here are the rules of each format. So notice we go into each format and then we say, these are the attributes that make this short form format compelling. So you have to get very, very specific about how you instruct AI to do what you're asking it to do. And this is why it's so. There's an irony about AI and technology, which is people think they need to become technology experts or AI experts. But the reality is, no, the technology is actually pretty easy to use. You need to learn how to have clarity over how you do the thing in the first place. And there is an art to speaking objectively and not subjectively, which leads to AI writing tip number six, positive attributes. So it's not enough to just get objective or say, this is what we're going to do, and I'm being very specific about it. But you want to reinforce the instructions with positive attributes. So you want to say, look, this is what it would look like for you to do this really, really well. This is where there's levels. There's levels to how you can instruct AI to create what you want it to create. So And I'll give you an example here inside of my short form Social content generator prompt is attributes that make this short form format compelling. So, yeah, you know, I want you to use concise language. I want you to say things in the most economical way possible. I want you to alternate between long and short sentences to add emphasis, rhythm and flow. I want you to use some sort of strong opinion, polarizing point of view. All of these things make sense, and they're probably things that you've thought of or heard of before. You can take this, though, to a completely different level by starting to really think about what other things have you learned or what other resources have you consumed where you could take little nuggets of information and plug them in here. So I'll give you a great example. One of the things that makes this format, this short form format, paragraph style work really well, at least in my content, is when you use these little mechanisms that are proven literary mechanisms. And there's an amazing book on this topic called the Elements of Eloquence by Mark Forsyth. And this is a whole book essentially about these different literary mechanisms that have been working since the days of Shakespeare. For example, alliteration. When someone writes something with alliteration, you sort of can't help but go, oh, I love the way that sounds, right? Or maybe you've never heard of this word before called a polyptotonic, which is a turn of phrase where you use the same word in two different parts of speech. So you use it as a noun and a verb. It's one of those things where once you hear it, you go, oh, that sounds so smart. But really it's just a literary mechanism or an antithesis. It's an obvious observation followed by a non obvious observation. So this is where AI is leveraged not just on skills, but on all of the knowledge that you've ever accumulated and you will go on to accumulate in your life. As I was building this prompt, I remembered that book that I had read. I remembered that I had highlighted a couple of those literary mechanisms and gone, oh, those are awesome. I should use those more in my writing. And all I have to do is explain that to AI and I say, hey, these are actually some of the positive attributes that make this specific style work so well. And so when you're thinking about listing out positive attributes, it doesn't have to just be things that you thought of on your own or your own knowledge. It can also be turning to other people's work and going, I really love how that person said this. I want to use that same mechanism in my writing here. And you almost bake it into the instructions. AI writing tip number seven, you want to use exact examples. So this is where training AI becomes a really important part of prompting. It's not enough to just create instructions. The way AI is really going to execute on those instructions is when you also give it examples. And where people go wrong is they will create. I mean, first of all, they'll make all of these mistakes in the instruction part. You know, they'll use subjective language instead of objective language, or they won't be super specific about what they're asking for. Right. Or they won't name each individual thing. But the real mistake that they make is they either don't give any examples or the examples that they give don't, as perfectly as possible don't match up with the instructions. So what I mean by that is, whatever thing you're instructing AI to do, you want your examples to mirror those instructions as much as possible. And when I say mirror, I mean down to the sentence or word level. So let me show you what this looks like in this format that I'm training AI on. What, how, why? I'm saying this format consists of three parts. I want you to start with a declarative opening line. This is the what I want you to follow with three to five bulleted examples, actions or steps. This is the how. And I want you to end with one sentence. Insight, benefit, or motivating reason. This is the why. Okay, so notice how these are the rules for this format. Now, the mistake would be saying, here are the rules for this format. And by the way, here are some examples of just viral posts that I found. If those posts in those examples don't match up with the rules that you you gave it, AI is going to get confused. It's going to say, wait, you're telling me these are the rules, but you're showing me examples that are not following those rules. So I guess I'm going to give you a blend of both. I guess you want more of the example. AI is not going to know what you're asking for. And so if you notice, that's why in this prompt, all of the examples that I give follow those rules down to the sentence level. I'm saying I want you to open with this one sentence. All of my examples open with that one sentence. I want you to do three to five bulleted examples in the middle. Every single example has three to five bulleted examples in the middle. As much as possible, you want your examples to mirror the rules. And said oppositely, the more your examples don't line up with the rules, no wonder AI is confused. And I find that of all the things when it comes to prompting, this is probably the one that writers get wrong most often is they either don't reinforce the rules or reinforce the training with examples, or they just pick random examples. And even they don't have clarity over why they're giving the examples that they do. So you always want to be thinking of the examples mirroring the attributes. AI rating tip number 8 Do not forget the delivery instructions. So what this means is, yes, you can explain. Here's what I want your help with, and here are the formats I want to train you on. And here are the positive attributes and I'm using objective language. And here are examples. All of these things are important, but ultimately it needs one final step, which is and how would you like the information presented? And so this is where, you know, I think, I think there's a lot of value in doing it at the beginning of a prompt and also as a reminder at the end of a prompt where you say, again, I would like these things presented in this way or presented in this order after I give you whatever it is I give you in this intro here. Again, I'm asking for short form social content, 280 characters or less. I'm going to give you the topic. I would like it presented back in these five time slots. I would like one short form piece inside of each time slot. And what am I going to give AI? I'm going to give it the topic. So being very clear about this at the beginning is key. And then you can reiterate it at the end. So notice here how at the end I say, okay, I'm going to give you the topic and I would like you to give me a day's worth of short forms, thought leadership, social content. Each post must read like a standalone tweet, without any formatting labels or headers, and must fit within 280 characters. Constraint of X Twitter. So you want to reinforce the delivery instructions at the beginning and at the end. And as much as possible, you want to say very explicitly, this is how I would like you to present the information. I'd like you to present these things in this order. I want it in paragraph style, or I want it in a bulleted list, or I want it organized by hierarchy. Hierarchy according to these rules, right? You can decide what you're asking for, but you want to tell AI, don't just go do all These things. Go do these things and then bring me back this specific output. And AI writing tip number nine is you want to and you can ask AI to iterate on your prompts with you. So what this means is even if you take the time to assemble a really long form automation prompt, you know, you write this long form, hey, I know how to automate this task. I want you to just do this over and over and over again. Help me create social content, help me write my weekly newsletter, whatever it is. The first or second or third or tenth time you run that prompt, you might realize that AI makes a mistake here or there. And so what you can do is you can ask AI, hey, why did you do that? Or what could I add back into this prompt in order to make it better? So this is a really in depth outreach prompt that I built for ghostwriters inside our premium Ghostwriting Academy. And basically what it does is it trains AI on how I as a ghostwriter would look up a potential lead and then how I would sort of assess and score what they're doing on social, what they're doing on their website, what they're doing via email. It helps me then become aware of where they're struggling and then I would know how to pitch them, how to pitch them my services. So I created a prompt that basically automates all of that for our ghostwriting students. So I'll show you how this prompt works and let's say I'm pitching myself. Okay, so I'm going to say Nicholas Cole, Premium Ghostwriting Academy. Because all in this prompt, all I have to do is give the name of the lead and their company, ChatGPT or Claude, can then look up that person online and then spit back a ton of information so you can see how this prompt works. Okay, so this is a great example. I notice that there's a couple mistakes in the first part of this output where it's pulling my social profile. So first of all, it says no branded channel found, but that's not true. I do have a YouTube channel and it's saying no secondary company found. But that's not true because we have Premium Ghost Ring Academy and we have ship 30 for 30. So when I notice these things, and this is how you tweak prompts over time, and this is sort of the whole art of prompt writing. What you can do is you can interact with AI and say, hey, I noticed some of these things. Can you double check this? Or what would you recommend adding back into the prompt so that this mistake doesn't Happen again. So let me show you what this would look like. Okay, great. I noticed though, in the social profiles section that you said Nicholas Cole doesn't have a YouTube channel, but I know he does. You also said there's no secondary company, but I know he has one. Ship 3430. How come you missed these pieces of information? And is there anything I can clarify in the prompt itself so that it doesn't happen again? So notice, like, you can interact with AI, you can tweak those things. So this is pretty cool. Notice how AI says, honestly, your original prompt was very well structured and detailed. However, if you're looking to bulletproof future runs, here are a couple ideas. So, one is emphasize brand name lookups alongside personal name lookups. So we could add this back into the prompt. Please search for, for YouTube specifically. If you're looking for any channel, then you could add this back into the prompt tweak. And then under company number two, okay, you could include a secondary company brand, newsletter, product, et cetera. So these are great little tweaks. So what would be the iteration process? You would run a prompt, you would see what breaks or see what isn't working as well as it could. You would ask AI, Hey, I noticed some of these things. What could I change in the prompt to not have that happen in the future? And then AI will give you those little tweaks, you add them back into the prompt, you run the prompt again, you see what happens, and this is how you iterate. And I think of all the skills, the iteration component might be the most valuable one because, yeah, it's amazing to sit down and just one take and like a 1500 word prompt. But it's very unlikely that the very first version of a person prompt is going to run perfectly. And so this is where there's this interesting intersection happening between writers and almost like engineers, where you want to think of yourself as a writing prompt engineer. You know, you're doing both. You're writing the instructions and then you're running them and you're seeing what the output is, and then you're iterating based on that output. One final thing I'll leave you with is I am a professional writer. I get paid to write. I have monetized my writing in all sorts of different ways. And I couldn't be more excited for AI. I think it is such a cool technology. I think it's going to change a lot. I think it's going to mean adopting new skills and new ways of thinking. But I am not one of those people who looks at it and goes, oh, this is going to ruin writing and the life of a writer is destroyed and there's no more jobs for writers. I am not in that camp at all. I think that AI is going to create 100 times more writers and I think it's going to make AI driven writers ten times more effective. I think it's just going to make all the opportunities flow to the writers who end up using and leveraging AI and all the writers who fight it and don't want to embrace it. They're going to really struggle. And that's just how it always is in every industry whenever some new innovation comes out. So I hope you enjoyed this. I hope this was helpful. If you have any follow up questions, drop them in the in the comments. I would love to keep riffing on this topic. I'm really passionate about AI. I use AI constantly. I'm always trying to learn more about it. I'd like to create more videos on this topic of Writing with AI. Subscribe to our newsletter Write with AI. And yeah, if you have any ideas for future videos, let me know because I love jamming on this topic.