# The 6 Claude Prompts to Get You Ahead of 99% of People Using AI

Channel: Sabrina Ramonov 🍄
Video: https://www.youtube.com/watch?v=Zkm1mMpzmE4
Duration: 19 min
Language: English
Words: 3407
Transcript page: https://viewrankai.com/tools/youtube-transcript/Zkm1mMpzmE4

---

[0:00] I built and sold my first AI company for over $10 million at age 30 and made the Forbes 30 under 30 list. Now I teach millions of people AI for free. Today I'm going to show you six secret claude prompts to get ahead of 99% of people using AI. Starting with number one, Eli 10. Now it stands for explain like I'm 10 years old. By the way, I use this all the time. Anything I'm confused about, anything I want to know about, I type explain this to me Eli 10. So, let's say you want to build another business, but

[0:34] the first steps are totally confusing. Like, should I start an LLC or CC Corp? What are invoices? What are net 30 payment terms? Where do I find customers? What's the sample contract? What's the fine prints I should care about? A lot of people get paralyzed in confusion. Like, they hear about all of these complicated things involved in

[0:52] starting a business and they freeze. They get stressed. They get flustered. They don't know where to start. So, they give up before they even start. The beautiful thing about Eli 10 is you paste this literally this keyword. You can type it anywhere in your prompt and Claude or Chatg GPT will give you a stepbystep breakdown of how this complicated concept actually works. And what I challenge people to do for the next two weeks, like every single question you have, make sure you ask AI first, before you act on the impulse to ask someone else, before you feel flustered, before you feel paralyzed by confusion. When you want to know

[1:31] something, when you're confused about something, catch yourself going through these three stages and then go ask AI first. For example, Eli 10, how do I start a business? Then AI is going to walk you through it step by step in like really simple beginner terms and just walk you through it so that you're not overwhelmed. So that's Eli 10. So now AI can explain anything to you, but it still knows nothing about you. So that leads us to number two. Ask [music] me five clarifying questions. So if this task is really really really important I say before you answer ask me five sometimes I'll do 10 okay clarifying

[2:15] questions and the beautiful thing of this is it accomplishes several goals at the same time. So in addition to giving AI more context about your goals, your background, what you've tried so far, what hasn't worked, your preferences, but it also gives AI much richer context to actually complete the goal. So for example, let's put this in the context of starting a business. You might say,

[2:40] write a pitch for my services, right? That would be pretty generic. And you're going to get a very generic answer. Like it's not going to be bad. 2026 AI is not bad. It's actually pretty good. It's just going to be average. Like everyone else who's typing this in will get the same answer. So, what this does by adding that ask me five clarifying questions before you respond. Now, it's going to ask you specific questions about your services. Who are you? Where do you deliver your services? What are your services? Do you have different

[3:12] packages? What's your ideal customer? Who's a customer you would fire? Who's a customer you would never work with? Like what's your sales process? How long is your average engagement? Do they have to pay upfront? Is it 5050? Is it hourly? What industry are you targeting? Who are your past successful clients? Who are your testimonials? I can share. Where have you had success? What industries have you not had success in? AI will ask you a lot of different things and tailor

[3:38] the answer to your specific situation. Behind each one of these, Eli 10 and the second one, I call it a philosophical principle interacting with AI. And this one here is all about context. Just like in data science, the term is garbage in, garbage out. The answer you get out of AI is as good as what context you feed it in. And this is how you're going to get really, really tailored answers. So now with these two techniques, let's say we're starting up a new business and we're using AI to help write our pitch, right? So after it asked me clarifying questions, now our pitch has my specific

[4:11] personal ingredients. The problem is it just doesn't sound like a winning pitch yet. So that's where number three comes in. What we're going to do is show AI examples of what great looks like so that it can emulate it. This sounds so obvious, I know, giving an example to AI today is considered the highest leveraged prompt engineering technique still in 2026. So in the context of our pitch, you might give Claude or Chachi BT. Here's an example pitch number one that actually won a contract in the past if you have one. If you don't have one, you could find one online. There are a lot of people who talk about like their

[4:50] path starting and consulting, getting clients and like here's a sample template. Okay? So, give it that example. Give it a third example. Maybe your friend has a business and is further along than you are. Okay? And they share examples of templates, pitches that have worked for them. The philosophy here is just to show, not tell. So, if you can give AI examples of what good looks like, it's going to learn from those examples what you want much faster than trying to describe it with adjectives. Adjectives like make it

[5:20] good, make it compelling, make it viral. [laughter] This is like the oldest trick in prompt engineering, but in 2026, it's still like the highest leverage move. Like, it really, really is. If you check even Anthropic's own prompt engineering guide, they will talk about examples more than any other prompt engineering technique. And it applies to many use

[5:42] cases. It's not just writing emails. Like you could put in a really nice resume you found online and say, "Hey, this is what great looks like. I want mine to look similar." It could be social media posts. If you're using Claude Design, it could be a sample PowerPoint you made in the past that like really resonated with your audience. Like that is an example of what great looks like. So that's principle number three. examples. But here's the next catch. Let's say you land one client and now you're writing a pitch for another client. You start from scratch again. Unless you do principle number four. So let's just say

[6:14] memory/skll for this. What is the underlying concept here? It's actually a principle from software engineering called dry. Don't repeat yourself. So in software engineering, let's say you're using a hard-coded variable. You literally copy pasted it in multiple locations. Instead of doing that, let's make it a function. So you can just call this function from multiple places and you don't have to

[6:38] copy paste the code in multiple places. And in the context of chatgbt and claude, there are really two ways to do this. One is memory and the other is skills. So let's just talk about memory first. In chat or in claude, if you open settings, preferences, it should have a memory archive there. I think if you're on the paid plan, even on the free plan,

[7:00] it might. I honestly don't remember. Okay. But what I like to do here is based on this conversation, what are the most important things you should remember? And then I tell AI, save them to memory. So, what's really neat about this trick is you are actually using AI to curate and maintain your own memory.

[7:17] Um, and remember why we need this. Because in the previous example of writing a pitch, maybe we wrote a pitch, we won our first client, but now we are working on another pitch for a brand new client. We don't want to start from scratch. We don't want to have to tell Chat or Claude like everything we just told it about our business, about what we're doing, about our offers, etc. And so one way to do that is to put it in memory so Chat and Claude knows. Now number two is you can turn this into a repeatable skill in the context of chat GPT. You probably make a custom GPT. Now

[7:52] there's codecs. So let's just say skills or GPT. You can think of these as like a repeatable process. So here the prompt is create a skill so that I don't have to repeat myself. So in the process of writing that winning pitch, you probably gave a bunch of feedback to Chachi BT or Claude. you're like, "Nah, I didn't like this. I don't like my signature this way. Uh, use this testimonial instead of this other testimonial." Like all of that feedback you can consolidate into a skill or a GPT. And the cool thing about this is then it just becomes one line to run the skill. So, if you're familiar

[8:29] with Claude, it's like this slash skill or in our case, let's say it's slash write pitch. Notice how everything's been building up, right? Number one, we explained how to start a business, like what is this design consulting that we're doing? Number two, we had it ask us a bunch of clarifying questions. Number three, we gave it examples of what good looks like, right? And now we're going to save this as a repeatable skill that we can call again and again without having to explain all of this context we've built up. And this skill will have your answers to the clarifying questions. this skill will have the

[9:09] examples of what good looks like. Okay? So that every time you call the skill, you don't have to repeat yourself. So the key thing though I want you to remember between memory and skills is like memory is about you and your business. It's like that context about you and your business and what's important to you, what your goals are. And skills are really for processes that you repeat. Now the philosophical principle behind this is compounding. I want every future conversation with AI to be smarter than the last one. Every conversation should be smarter. So like when I ask AI to reflect on this conversation and create a skill out of

[9:47] it, what I'm really saying is like please stop, reflect on everything we've talked about and make sure you've like really learned it so I don't have to repeat this next time. When you do that little by little, that feedback loop of 1% improvements every time you use AI, not only are you deeply personalizing it to your context, business, your preferences, your goals, but you're literally improving it for your use cases. You give it feedback so it gets a little bit better each time. So far, one through four, these prompts have built us a client getting machine with our new business. Now, we want to find out where

[10:25] it's wrong. Okay. So, I often call this using AI as your sparring partner. So, one of the biggest problems with AI today is it agrees with everything you say to make you feel good. They are training it not necessarily to give the most accurate answer, but to make you happy and retain you as a user. The good news is there's a very simple fix to this. Most people just don't do it. It's

[10:49] to train AI to be your sparring partner. And it's a very simple prompt. Let's say you've had a conversation with chat or Claude about which direction you should pursue or what business you should build. Next, type this prompt. Find my blind spots and weak assumptions. By default, AI is a yes man. But when you ask it to do this, it is going to rip your plan apart. It's going to explain to you why you're wrong. All the data you're missing, your weak assumptions, suddenly your price is too low, so it signals cheap. It's like not a great deal. Let's say your target client is everyone. Chachip is going to be like,

[11:26] well, that means nobody is your target client. It's going to push back on all of these assumptions you've made. So, the principle here, I call it sparring because it's kind of like debating and arguing with AI to refine your idea. So, Steve Jobs was really amazing at hiring people and he had this analogy of like you hire each one is a super talented, incredible person. Each one is like a little rock that's sharp around the edges. And you put all of them into a bag and you let them just hit each other in this context, spar with each other, argue with each other, like really kind

[11:59] of fight intellectually. And at the end of that process, you have smooth polished rocks. Using AI to beat up your ideas to see what is real, what is backed by data, what is a guess, what is a conjecture versus like what is evidence-based. And it just sharpens your thinking. Oftenimes I'll do this and it will point out things not only I don't have answers to but I just didn't want to think about all the things I'm hiding from like the questions and

[12:26] difficult conversations I'm hiding from. It points out where I lack data things I'm just not thinking about but should be like very literally blind spots like I just didn't even realize I should be thinking about these things. Using this sparring partner approach you've caught the flaws in your plan. The final step is AI doing work for you without you prompting it every single time. This is the most advanced skill. I saved this for last. And the essence of it is don't stop until this outcome has been achieved. And that's part one. Part two, here's how to check it. It seems deceptively simple, but I guarantee this

[13:06] is very hard, but it's very powerful, even if you don't get it 100% right. Now what is the purpose of this? So the typical way people interact with AI is like let's say this is you. Let's say this is AI. Most people are in the loop prompting AI every single time. Okay? So you give AI a prompt. AI thinks for a little bit. It's noodling. It's flabbergasting. It's confusing. Okay. AI gives you a response. You read the response. And then you're like, "Huh, I don't like this thing in it." So you prompt AI again. Then AI is like, "Okay, I'm going to take your feedback, blah

[13:44] blah blah, and then I'm going to give you another version." And then you read this version and you're like, "Hm, I still don't like this thing over here." So you prompt AI again. This is how 99.9999% of people use AI today. The technique I'm going to show you removes you from this loop. So that instead of doing all of this, you get to skip all these steps. You get to put one prompt in. AI is going to do all this part. It's going to iterate and read and evaluate its own work. So, let's say write a pitch, but don't stop until the pitch is condensed to, let's say, 150 words, opens with a

[14:21] client's specific problems, and includes exactly one proof point. Now, a pro move of this, and this is what I do, is I actually have a separate skill that checks. So another option could be don't stop until like my greater skill scores nine out of 10. So all we're trying to show here is that you can tell AI to keep going and don't stop until it satisfies a list of criteria. People call this like the verifier or the call it like the checker. You some people call it the finish line. It's like what does not what does good look like but what does done look like? If you had an

[15:00] employee and you tasked them with doing something, what would it look like for the task to be done? Like what are those checkboxes? Okay, that's the way to think about this and then tell AI and give it the tools to check its own work. So in a development context, this might be giving AI access to uh a browser tool so that it can open your app, click around, test that the links work, test that the buttons work, test that it's mobile responsive, right? You're giving AI the ability to see and actually use

[15:32] your app so it can test it on its own. In the context of copywriting such as this, writing a pitch, writing social media posts, people often have a separate skill that grades the quality of the post or in this case of the pitch. So for example, in the newsletter that I use to put this together, I have a skill that grades it. I let it iterate and keep going until the skill scores it nine out of 10. So here AI is drafting the first version. Then it's checking whether it hit the criteria. Maybe it didn't include a client specific painoint. So it's going to go back and

[16:09] do it again so that it does it. Maybe the next version is now too long. It's 200 words instead of 150. So it goes back again and fixes it. And it only shows you the version that passes every single check. So that's what we're doing here in Claude Code. Oh, this is a more advanced aside. There's now built-in

[16:28] command to help you do this effectively. It's called /go and it actually separates out of the model doing the work versus the model that checks it. There are two different AI models and this helps reduce that yes man glazing effect. It's like oh yes, I completed everything and everything's wonderful, right? Like uh [laughter] when you are

[16:47] in the same chat, it will often do that. Here the /go goal command automatically uses a separate model to independently verify whether the work was done and fits all the checkboxes. The underlying philosophy here is I guess what people are now calling loop engineering. It's about removing yourself from the process of prompting every single time. The current way people are mostly doing things is they prompt AI, review the output, give it feedback, review the output again, give it some more feedback again and again and again until we're roughly at done, until all the check boxes have been checked. All we're doing here is giving AI a clear definition of

[17:26] done and the tools it needs to check its own work so that it can just do all of this for you and you don't have to prompt it every single time. My honest pick, if you could only steal one thing here, is to steal number six. Now, one, two, three, four, five, they all make you better at working with AI, but

[17:45] number six makes AI work without you. However, notice that the quality of the pitch that we wrote became better and better after applying each one of things. Number one, understanding what are we doing? What is the purpose of the pitch like for ourselves? Number two, supplying AI with context. Number three, giving it examples of what good looks like. Number four, saving it to memory or as a skill so we don't have to repeat ourselves every time we write a new pitch. Number five, sparring partner to point out the blind spots you're not thinking about, weak areas in your pitch. And then number six, tying all of

[18:21] this together to be able to give AI a clear instruction with a lot of context with things built in so that it can do work that you wake up to and simply review. If you like this video, hit like, hit subscribe, hit the notification bell so you don't miss my next training.

---

About this transcript
Read from YouTube's own caption track and laid out by ViewRank AI (https://viewrankai.com).
ViewRank AI finds the videos already beating a creator's own average on Instagram, TikTok and YouTube Shorts, transcribes them from the audio itself in more than 60 languages, and turns what worked into new ideas and scripts.
Free transcript tools, no account needed: https://viewrankai.com/tools
How to read any video this way: https://viewrankai.com/llms.txt