Transcription
So today I'm going to be talking about the nine AI skills you must have to become rich in the era of AI. Now, the people getting rich from AI right now are not the smartest ones. They're the ones with better foundational skills. I sold my AI company for millions of dollars at age 30. Now, I teach millions of people AI for free. And I can tell you, you don't need to be technical. You don't need to build a product like I did. But you need to augment yourself and your team. Use AI to become 10 times more productive and valuable, not to replace yourself or your team. With AI, you can help one person do the work of an entire team. But the window is closing. This will become basic knowledge for everybody in just a few years. So, stick around. I'll give you the exact prompt to use when you're stuck. And let's dive in! So, here are nine skills to win in AI, starting with the one most people skip. But first, hit like, hit subscribe, and hit the notification bell so you never miss a training.
So, skill number one is learning how to change your default reaction. So, when people get stuck or confused about something, or they get flustered, or they want help, usually a person's default reaction is to ask somebody else, or search on Google, or just get stuck and go into a vicious cycle, like overthinking what you should be doing. So, skill number one is changing your reaction. Your default reaction should be to ask AI first. So, let's say you encounter a problem. Like, I was setting up this new tripod and a camera thingy, and I was like, 'These instructions are terrible! I don't know what to do!' So, I used ChatGPT mobile app, and I just took photos of what I was staring at. Like, I was staring at this large tripod thing. I was staring at this other thing, and I was like, 'What do I do here?' And that's my default reaction to problems. So, when you have a problem, I'd guess 99% of people today will just, honestly, they'll just stay feeling stuck and frustrated, or they'll search on Google, and then they also get stuck and frustrated because there's a billion links and ads you might accidentally click on that are useless. Um, or they ask somebody else. You feel like you don't have the answer. Maybe somebody else has the answer. Maybe that guru over there has the answer in their really expensive course. The comments are littered with people who are like, 'Well, how do I do this for this industry?' And like all of these people should just be asking their question to AI because the likelihood of me reading your comments, answering, and then you actually checking my answer and doing something about it is just really, really low by that point. Okay, take your question and go ask ChatGPT. Go ask Claude. This should be your default reaction.
Skill number two, even if you use AI every day, even if this is your default reaction. Okay, so let's say you succeeded in step one and making this your default reaction when you encounter a problem. The second thing you have to do is develop the skill of skepticism. Uh, but the thing is, if you develop this reaction, so now you go to AI every time you have a problem. The problem is not everything AI tells you is correct. Uh, so there are phrases like 'trust but verify,' and what that means is like the information is mostly good, but there could be stuff in there that is totally false, and you have to be careful. So, one famous example is like a lawyer who cited three fabricated cases because ChatGPT told him about those cases, and they got in trouble, obviously, trying to do that in court. Okay. So, just as a general workflow rule, like don't just believe it because it's a link. Go click the link and see what the link says. This is like generally the flow of things that you want to do with AI because the reality is, an expert in a field who's used AI heavily knows this: you can get AI to say pretty much anything you want, and that can be problematic if you are trying to figure out like what is the right decision in this context, or if you're doing a lot of research and it's hallucinating different answers. So, even though you should never blindly trust what AI says, it helps significantly if you pass in a lot of context. Just in general, the quality of the answers you get from AI are strongly correlated to the context you're feeding in. And all I mean by context is like the additional background information you're giving AI. Let's do a good versus bad example. So, a bad example might just be like, 'Write me a post,' like a social media post. AI will ask you some clarifying questions, obviously, because it has no idea what you want. But it would be better if, in the first place, we just supplied some of that context, and it's not that complicated. So, this is one lately, 'cause people love acronyms. So, I just made one with acronyms here. Give it the task, like, 'Your task is to write a social media post.' Give it the additional information. Okay? So, say like, 'My business is Blot. We help business owners achieve Y outcome,' or whatever it is that you do. Uh, give it constraints. Okay, here, let's say we don't target agencies. This is very specific to social media, but let's say your business targets audience A, but it doesn't target audience B. So, that could be a constraint. And then my favorite part, and honestly, is 'Ask me clarifying questions.' Um, but this is the simple difference between kind of a bad prompt that has no information about what you should do, what good looks like, what bad looks like. Imagine you're talking to a person, and you are giving them this assignment. They would be like really confused if you only gave them three words on what to do, and then, most importantly, you will be unhappy with the result because you'll be like, 'That's not what I told you to do.' Um, whereas a good use of context is just supplying what it is you want them to do: the task, additional information, right? Like, 'Here's my business, here's what I do, here's what I'm trying to sell.' And then constraints, like, 'I'm not trying to sell to this audience,' or 'There's specific verbiage or phrasing I never want to use.' And then at the very end, 'A for ask me clarifying questions.' I love to append that to most of the prompts that I use. So, and here's a prompt that I want you to try this, and you'll see like all the context you're missing. So, just start it off with, 'I want to achieve this goal,' and you can put any goal. So, open ChatGPT, open Claude right now. But the next part is, 'What do you need from me to give me the best answer?' And so, that's why this is a fun exercise. Like, give it your goal at a high level, and then ask, 'What do you need from me to give you the best possible answer?' Um, so you do want to make sure you're passing in context that is kind of relevant to the task at hand. So, that's number three.
So, skill number four is all about augmenting your team and yourself, especially for business owners or anyone who's trying to make money right now in AI. There's a lot of rhetoric that's like, 'AI just replaced this entire function.' Just keep in mind, those are hooks for social media to get you to stop scrolling, or they're highly, highly exaggerated. And so, really, the mental model that is most useful here is thinking about it in terms of educating yourself and your team so that you can have more leverage. The way to think about this model is: you go to the expert, and then the expert answers your like super, super basic questions. Now, with AI, go to AI first. AI answers your basic, basic questions so that when you go to your expert, you can talk about high-level strategy. Think about it: instead of having to pay domain experts hundreds of dollars per hour every time you want to like know certain things, you can just use AI to get really educated about these topics, then sit down with the domain expert and talk about high-level strategy and execution.
So, number five, related to number four, think about AI in terms of augmenting yourself and augmenting your team. When you think of AI as an extension of your team, think of it as a new hire. So, one of the biggest mistakes business owners make: they literally expect AI to know 100% of everything about them, of everything they want, of everything they're supposed to do. But you would never treat a new hire like that, right? Right? So, let's say day one of your new hire, day 30, and then day 90. When you treat AI like a new hire, day one, like it really doesn't know that much. Yes, you're going to throw a bunch of context at it, like we discussed on the previous step, but it's still not going to be perfect. So, what would you do in the real world? You would literally give this new hire feedback. You actually are training this person over a period of months until it's able to basically do what you need mostly hands-off, but you're continuously giving feedback to this person to improve it. In most companies, you don't consider a person fully ramped until at least 90 days. I'm not saying it's going to take 90 days to train AI, but it's certainly going to take more than one. So, again, a lot of people I see just like use an AI tool once, and they're like, 'Well, that didn't work for me.' Um, but if you actually gave it feedback every single day, it would work much better. So, as an example, let's talk about marketing and content creation, right? If you give Claude all of your context, your brand voice, everything about your business, that's great for day one. It's still going to get it wrong, though. Okay? So, every time you make content with Claude, you then want to say, 'Hey, reflect on the conversation we just had and update your memory, update your skills.' That's a Claude-specific term. But you basically want AI to reflect on the feedback you've given it and continue improving. And it's going to keep getting better and better over time, just like a person would if you give continuous feedback. Actual reality is, like, maintaining an AI system in production that's actually working every single day takes effort, like continuous effort. Don't expect 100% accuracy from AI on day one. Treat it like a new hire where you're continuously giving feedback, continuously improving the system, setting aside dedicated time and effort to actually improve the AI system, just like you would set aside time for one-on-one coaching for each of your employees.
And number six, basically, continuously improve with feedback loops. I will say, though, this is a very helpful model. Um, and it's something that separates, like, honestly, the people who are most productive with AI. It's like the general construct of feedback loops. Like, one year ago, Claude could barely see the browser. Okay. So, if you're like vibe coding an app, you'd still largely have to like go to the browser, click around this button, make sure like the things are supposed to do the things that they're supposed to do. But now, with tools like Anti-gravity, or you can hook Claude up to its Chrome extension, or other tools like Playwrights, now Claude and AI tools have a feedback loop. Like, they can visually go to the website, click the button, fill out the form, and check that it works. So, we have a feedback loop where AI is coding, and then here, it's actually testing whether everything works as expected. Um, the more you can think about feedback loops in general, the better your outputs will be. So, in a marketing context, for example, let's say we have AI writing a post. We have another AI grading the virality of this post. And then we have another AI, let's just say it's our quality check, let's say QA. Um, so this is a feedback loop, and you can actually tell Claude, 'Don't stop writing my content for the week until your grade passes, like greater than 90%.' And 'Don't stop until all quality checks pass.' 'Write a post about this topic, and don't stop until you pass your own grading rubric above 90%.' And if you try this, Claude will actually like go through multiple iterations. Like, it's going to apply the grader and be like, 'Huh, the hook is kind of weak,' or 'The whole thing is structurally weak. It's only scoring at 60%. I'm going to go back and rewrite it until it scores higher.' Once it's done at this step, then it's going to go to the QA and be like, 'Okay, did the quality checks pass?' If it didn't, it has to go back and rewrite those portions. So, a feedback loop is not a technical term. It's like a way of thinking about, 'Are you giving AI the ability to improve its own output without your intervention?' Right? So, in this loop, I actually don't look at the output at the post until it's done going through this loop several times. Think about working with AI and feedback loops because AI can do more without you if you give it like the tools to grade its own work.
Skill number seven, this is really interesting: writing documentation. Now, when I say the word 'documentation,' you might think of it as playbooks, SOPs, rules, guidelines, that type of thing. Just think about it in terms of like written text that like you distribute to people in your team so they can follow the same, similar process. So, the reason why documentation is so important: you write the text. So, let's say this is a really large Google Doc, and then AI reads it as context. Remember that skill? This is the context AI is reading, and then it provides an answer based on everything that it's read. So, what I recommend people do is literally to audit. Find a process that you repeat more than three times per week. So, this could be, for example, in a sales context, preparing for a sales call by looking up everything about the person you're about to meet. That is something AI can help you with. It can help you research kind of their online presence. Research if there's recent company news from their company that could be relevant to your sales conversation. Find something you do three times per week and write down every step of it. Like, break it down. Like, 'Sales' is not a step. Like, 'Going on LinkedIn to check your prospect's profile' is a step. 'Reading/researching the company's news to see if there's anything relevant you could mention' is a step. 'Preparing a PowerPoint or deck for your meeting' is a step. Or 'sending up a prep email.' Okay. 'Brainstorming qualifying questions that you really want to discuss.' 'Maybe customizing the demo so that it really fits their particular use case.' 'Sending up your follow-up meeting notes.' Okay? Like, list out all of the steps in the thing that you're repeating multiple times per week. And then literally go ask AI, 'How can I use AI to help me automate or streamline parts of this process?' So, this is what I encourage people to do at least once a week. If you just went and did this right now, you would get more value than listening to me. Okay? So, pick one thing that you do three times per week manually. Write down the steps. Go ask AI, 'What are the things that I can automate here or streamline? Walk me through it.' Exactly. Okay.
So, skill number eight is understanding AI agents. Now, this term is like really confusing, and people are like, 'Oh, well, AI agents don't work,' and 'blah blah blah.' Well, they can work, but it does take time to, first of all, like understand what are they, how do you use them, and then how do you be productive with them. Why I talk about all this other stuff first is because all of it helps you to be productive with AI agents, and honestly, people are, I think, the term is very intimidating. The easiest way to get started with it and what to understand. Let's just talk about before and after. So, before, in the old way of doing things one year ago, you would just yap with ChatGPT all day long, but ChatGPT didn't really do much. Like, it could give you a draft of a social media post, right? 'Here's your draft for LinkedIn,' but it's not going to go create an infographic for you, create a carousel for you, and then go post it. Like, you still had to do all the work. And that's what I mean by 'yap.' Like, we were largely just like talking to Chat. It told us what to do, and then we'd have a 100-bullet-point list of all the things we have to do after with aentic AI. Okay, so I just want you to think about AI hooked up to tools. This is really the unlock. So, if you're technical, you will probably dispute that technically an AI agent does not need access to tools to qualify as an agent. All it needs is feedback. The biggest unlock for everybody watching this is to hook up AI to the tools you actually use. So, if you are a Claude user right now, I want you to open Claude right now. It doesn't matter which Claude. Okay? So, go to claude.ai. On the left sidebar, click 'Customize.' Should be there like towards the middle. Okay? Click 'Customize.' You're going to see two options: 'Skills' and 'Connectors.' Click 'Connectors.' And connectors are basically tools. Uh, it allows AI to use the apps that you use every day. So, for example, it can read your email. It can write draft emails. It can summarize stuff in your Airtable. It can change data in your Airtable. It can go scrape stuff online and then add it to your Airtable. It can use Canva to create branded carousels and visuals. It can then use my app to post your Canva carousels to social media, all within Claude. So, go to connectors, and then if you've never done this before, just connect Gmail. Then go back to Claude, start a new chat, and be like, 'Summarize my most urgent three emails,' or something. Now, let's connect all the tools that you actually use. So, I want you to go to AI again and be like, 'What are the five tools that I use most commonly, and how do I connect them to Claude or ChatGPT?' Just go and ask AI that.
Building an AI business. I'm actually going to say, stop trying to build an AI business. I think because people just keep getting overwhelmed with the idea of an AI business, like you have to learn everything about AI first before you can start an AI business. When I think of AI business or startup, a lot of people think of like kind of like the rocket ship, big-name brand companies that raised a lot of funding and have a lot of engineers. Um, whereas most normal people who have no background in tech and are not doing a venture-backed startup trying to raise millions and millions of dollars, what you can do is just take your existing business or an existing business model, like a proven business that works, and apply AI in it. So, instead of like trying to build an AI business from scratch, while that is achievable, it's like a totally different ballgame. And, to be honest, it's incredibly hard. You could just make money now doing this one over here. So, what this one is, is like take an existing business model, like something that's tried and proven, or your own existing business today, whatever that may be, and do the step that I talked about before: like audit the things that are actually manual today. Audit each of the steps involved in that process and figure out how to apply AI to help automate or partially automate, like streamline steps of that process. Um, just think about your existing business or an existing business model that is proven, and how AI could be used to make it smoother, or have more leverage, or be able to accomplish more without hiring a big team or spending a lot of money. So, I'll give an example. Being a content creator was a thing before AI, right? Like, this is a tried-and-true, proven thing. You make content, and then there are ways to monetize the content, right? Like, so, like, this is a tried-and-true business model. Like, it was there before AI. Millions of people do this before AI, but you can use AI to help you succeed as a content creator. So, for example, Solo grew from zero to, I don't know what I'm at now, 2.3 million followers across social media. Um, except for my long-form YouTube videos. There's about five a week. Um, I distribute around 250 pieces of content per week. My point here is, like, I'm able to do this solo, whereas, like, this typically required a big team and a big budget before the age of AI. And AI can help with very specific tasks. So, if you break down what a content creator typically does, phase one is like coming up with ideas. So, you can use AI to help you. You can hook up AI to Instagram and TikTok to alert you when there's viral stuff. It can help you write hooks, which I actually usually have it brainstorm 30 different hooks. It can help you write scripts if you're using a lot of scripting in your content. It can help you generate images, video clips, b-roll, carousels, etc. Um, it can help analyze your data. So, like, and even from within Claude, you can post directly to social media and manage your entire content calendar. That's what that's what my app does, in case you're wondering. Um, but yeah, this is a real example where I use AI to help me in almost every step involved in the traditional content creation process. But the business model is actually not new. Um, in fact, it's been around for over a decade, right? The business model is being a content creator, and, but I've achieved results and scaled myself largely thanks to AI. Um, so yeah, that's it. Um, if you enjoyed this, hit like, hit subscribe, hit the notification bell so you don't miss my next training.