Transcription
So, agentic AI is on its way to replace around 50% of all jobs. Tools like Claude Coder are being used to do around 60% of the tasks in companies. And if you don't know how to use it, you might end up like the hundreds of thousands who have lost jobs because of AI.
But this is simpler than you think. You don't need to learn how to code in order to stay ahead of this. You need to learn one skill, and it's a skill that anyone can pick up. I have thousands of members in my community who have mastered it in just a few weeks, and none of them had any technical background.
So, in this video, I'm going to show you what's really happening with these jobs, why this AI is so powerful, and the simple steps that you can follow to learn how to use it. So, let's get into it.
Now, when you hear AI is going to take jobs, you probably picture a robot rolling in and replacing you. But that's not really how this works. What's actually replacing you is just a human. And I'll let the CEO of the biggest tech company explain this to you.
"Um actively and aggressively, you're doing it wrong. You're not going to lose your job to AI. You're going to lose your job to somebody who uses AI. Your company is not going to go out of business because of AI. Your company is going to go out of business because another company used AI."
Basically, what happens is that the person next to you learns to use AI. And they start doing your job and their job at the same time. So, the company doesn't need both of you anymore.
Now, why is this happening now and not 5 years ago? Because the AI we have today is different. Old automation only ever did exactly what you told it. And if something went wrong, you'd have to fix it. But the new AI is what people are calling agentic. Agentic just means the AI can take action on its own towards a goal instead of only answering your questions. So, instead of you doing every step, you just tell it the outcome you want, and it will go figure out all the steps, and it will do them for you.
And it's not just me saying that this is a big deal. Agents are coming for the kind of work people do at a desk all day. So, here's how to prepare for this. And it's actually really simple. If you can't beat them, join them. You're not going to stop this from happening. There's just no way. But you do get to choose which side of it that you're going to be on, either the person holding the controls or the person who might lose their job to AI. And once again, the good news is you don't need to know how to code or be a super technical person in order to do this.
I'm going to show you how to use it in less than 15 minutes. But I'm actually going to show you what you'll be able to do. So, let me show you three jobs that it can already do today. So, let me show you guys a few practical examples here.
The first thing I want to call out is I'm using the Claude desktop app. And if you earlier when I said Claude code, you got a little scared, there's no reason to be. This is Claude chat, right? You can come in here, you can talk to Claude, and it's basically just using your natural language to get what you want. Except for in the Claude chat, we're missing the whole agentic capabilities. We're missing the ability for Claude to really feel like an officer at your company and a full employee, rather than just a helpful little tool.
So, all we have to do is switch over from Claude chat to Claude code. And this is the interface. You know, we've got our ability down here to talk. The only difference is now we're working out of local folders and files. So, if I pull up my file explorer, and you can see right here I've got my camera roll open. You can see I can go to my desktop or my downloads. This is all we're doing is we're working with Claude, but instead of on the web, we're working inside of our actual local files. So, that means it can look through my pictures, it can look through Excel sheets and Word docs that it creates for me, and it can help me organize those, move them, create more, all that kind of stuff. So, it's just way more powerful. And it has a much better memory than just using Claude chat. Even if you're using chat and you're organizing stuff by projects, and you've got, you know, a few connectors plugged in, this is way different because like I said, it's looking at all of our files.
So, real quick, if I just said, "Hey, can you quickly tell me who I am, what my business does, and you know, what are some of our goals for this year?" Because my Herc 2 project, which is kind of like my AI operating system is what I call it, it can read all my emails, it can see all of my communication, it can look through all of my meeting transcripts. It knows everything about me and my business, as you can see right here with what it is currently spitting out. It knows my YouTube subs, it knows, you know, our two different communities in school, it knows our certification program we're working on, it knows my book, it knows my team, it knows all of this kind of stuff.
Anyways, the three examples. This first one, take a look at this. So, I came in here and I did something called a slash goal, which means I'm able to set a goal, and Claude will keep working until the condition is met, which is super cool. I mean, how much more agentic do you get than that? So, this is the prompt I I shot off. I'm not going to read this whole thing, but feel free to take a look at that if you want. I basically wanted it to pull in all my videos from quarter two of 2026, analyze it, comments, click-through rate, all the stats, and help me look at those insights and tell me what to do about it.
So, as you start to go through here, what you'll notice is this is not technical at all. This is basically just Claude thinking about what to do and looking through sources. So, it said, "I'll start by exploring what YouTube data and tooling is already available. Then I'll figure out how to pull Q2 analytics." So, it read a markdown file that I have called youtubechannel.md. So, it reads the goal and it needs to understand exactly what it can pull. So, it starts looking through things, right? It uses this Python scripts to actually be able to get the data from my YouTube channel. It has the right token. It continues to take action and then reason, and take action and then reason. And it just goes all the way through until everything's done. You can even see here, once it's actually created an Excel sheet, it screenshots it because it's supposed to verify it, right? It's going to make sure the colors appear. It's going to make sure the spacing's right. And it's not going to hand me something until it actually feels confident that I'm going to like what I got.
So, let me pull up the full Excel sheet so I can show you guys what I just did. And just again to prove my point, inside of my Herc 2 project, it created all of this stuff. And here is the actual Excel sheet that it made for me locally. So, I opened this up, and this is what we got right here. If we go to the first tab, what we can see is we have an assessment. We can see the different tabs like the executive dashboard, the video scorecard. We can see the views numbers. We can see the different data sources. We can see all this kind of stuff. I can go to my executive dashboard, which shows me from April through June I've got these amount of views, this much watch time, these many new subscribers. And we can see some of the most important things that matter. Claude code is the engine, massive subscriber quarter, and some bad things. Impressions and CTR were missing. So, for some reason it wasn't able to pull those, so we would be able to work that back in. Topic concentration is a risk. So, as you can see, not only is it pulling data, but it's helping me analyze it a little bit.
As we keep going through, we can see a video scorecard. So, I can see some of the videos that I've uploaded this quarter, and it's going to tell me, based on the views and the length and the viewer duration and all this kind of stuff, which ones are good. It's also putting them in different pillars. So, we've got Claude code and Agentic. We've got Voice AI. We've got Other. We've got Selling. All this kind of stuff. I can go through my monthly trends. I can go through content pillars, format and length. I can get into the audience and traffic. So, all of this stuff is going to help me every single quarter do a review like this, analyze the insights, analyze the data, and then help me work on my strategy for the next quarter of YouTube videos. And I want you just to real quick think about how much data that was, and how long would that have taken me manually to go to YouTube, extract all of that, put all of that in an Excel sheet, and then analyze all of it. As you can see, I uploaded 75 videos in this quarter. So, that would have taken me at least half of the day to just write all of this down to just pull all the data in here, get the spreadsheet looking nice, and analyze it. Whereas Claude was able to do this in about 10 minutes for me.
Okay, let's look at example number two. I once again utilized a {slash} goal prompt so that I could set the condition and it would keep working. This time, I wanted to pretend that I had a small cleaning business, and I wanted to build myself an app so that I could track my jobs and how much I'm getting paid and all that kind of stuff. Once again, same thing as last time, it reasons through, reads the instructions, it decides what to do, it makes a plan, and then it starts reading files, and it starts executing. And every time it executes, it reasons and then executes, and then it reasons and then executes. And that's basically the loop.
You can see here, this time what it created for me was an HTML file. So, let me go ahead and open up this HTML, and here's what we have. Cleaning jobs, we have today's date, we have how much we made this month, how much we're still owed. You can see, let's say Sarah Lynn paid me. I could go ahead and mark as paid and then mark done. If Mike goes ahead and pays me, I could mark paid and then mark that as done. Let's say for some reason Amy Brooks cancels, I can just X that out and delete that job. And I can also come in here and add jobs. And now, the cool thing is, we asked for this app to have a memory. So, basically that means if I was to refresh this, all of that stuff that we've done is going to save. It's not just going to reset every time.
Now, think about this. Do you know how to build this? Could you manually write this HTML? Nope, I couldn't either. But, I know how to explain to Claude what I want. If I want this to be different colors, I just say that. If I want there to be more functionality, I just say that. If I want to turn this from an HTML into an actual URL, an actual website that I could pull up on my phone, or if I wanted to turn this into an iOS app or a desktop app, I would just ask Claude to do that for me. And once again, this took minutes.
And finally, this third example, lead generation. How many people sit there and not only struggle with lead generation, but some people's full-time job is lead generation and outreach. So, look at this. So, for this goal, I had it look in Clay to find me 50 leads that are my exact avatar. Not only find those leads, but enrich them. So, find out what their business does, their business pain points, Google reviews, things like that. And then help me, based on the data knows about me and my business, create outreach messages for all 50 of those leads.
Once again, it reasons, it executes, it reasons, it executes, it reasons, it executes. And what it did for me here is it created this Excel sheet right here. Leads, July 7th, 2026, which is today's current date at the time of filming. So, I went ahead and opened up the Excel sheet, and what do I get? This Excel sheet. I've got 50 leads here. If I scroll through, you can see that we have 50 columns of actual leads, or sorry, rows. And the columns are business, decision maker, title, email, email verified, prior email status, phone, website, location, Google rating, review count, business pain points, recent current signal, notable achievements, personalization hook, hook source, email subject, email body, and needs review. So, all of this.
Think about how long that would take you as a human to go find, enrich, and write outreach messages for 50 leads, whereas this took me about 20 minutes by just asking Claude code with my natural language. And I'm not a master, you know, cold email copywriter, but just take a look at this. It's not very bad, right? So, the business pain point is that a recent one-star review calls out chaotic communication. And so, if we go over here to the email subject, your one bad review is about callbacks. And then the body says, "Hi Ante, saw your review about a customer still waiting on a callback a month later. A few different people reaching out with no one following through. It's a rare miss for a shop at 4.7 stars across 41,000 reviews. We recently helped a local business turn calls that it was missing into book jobs without adding anyone to the phones. So, honest offer, I'm still getting this off the ground, no case study yet, but I'll set it up on your line, run it for free for 30 days, and if it doesn't book you jobs, then you owe nothing."
Now, here's something to think about. Claude Chat could connect to Clay and could find you leads and could enrich those leads, but Claude Code knows more about my business. It knows our case studies, it knows our avatar, and because I'm building my AI operating system with its own second brain, using Claude to me feels like I have a co-founder rather than just, you know, hiring a virtual assistant, which is what people used to compare AI to a lot. It was like hiring a VA. But now, you can literally have a founder at your business, or you can have an officer at your company. And let's say you don't own a business, that's completely fine. If you're an employee, you now are able to manage a bunch of different super high-level employees, super high-functioning and high-performing employees, which makes you look 10 times better. Because once again, you can do 10 times more work and keep the same level of quality than other people who are in your class or, you know, have the same job as you.
So, if you've never opened up Claude Code because you were scared it might feel technical or because you thought it didn't relate to you in your position, this is your wake-up call. Please, please give it a shot.
All right, now here's the thing all three of those examples had in common. Every single one of them was a job that a real person gets paid to do today, or at least used to get paid to do. And now, any non-technical person can do all of that with Claude. Now, I'm going to show you how to use it in three simple steps, and we're going to do all of this with Claude Code, because right now I think it's the best one to learn.
But real quick, remember at the beginning of the video I said you only had to learn one skill. That skill is being an AI manager. Think about how you might manage a new hire. You would onboard them. You'd let them get to know you and get to know the business and get to know their role. And you wouldn't dump everything on them in week one. You'd slowly phase them in. You'd explain exactly what they should do, and you'd explain exactly what they shouldn't do. And then you'd watch them until they prove that they can work without mistakes. And when they start handing you work back, you don't just accept it blindly. You review it. You give feedback, and you help them get better. That's exactly the mindset you should have when you're working with AI. You're not the engineer or the operator anymore, you are now the manager. Your whole job is to make this thing as good as possible. And the three steps that I'm about to give you are basically just that onboarding plan.
Okay, so step one, just start talking to it. Open up Claude Code and treat it like a really smart person that you're handing a task to. Explain what the goal is, what a good result looks like, what a bad result looks like, and what it should avoid. And practice that on small stuff. And this is the onboarding part. Let it get to know you. Tell it what you do, how you like things done, what your week actually looks like. The more it knows, the better every single answer is going to be.
Okay, step two. Pick one real thing that you already do and try to do it with Claude. I mean, grab something from your week that you actually do over and over. And it doesn't even have to be something that's associated with your work. Maybe you just want to have AI help you plan your gym schedule and your meal prep or figure out your groceries. The most important thing is that you pick something that you will actually use because there's a huge difference between feeling the ROI yourself and just watching demos of other people talking about all the stuff that AI can do. You're only going to get hooked on it when it actually saves you time on something that you really care about. So, truly, the goal of step two is to convince yourself as fast as possible that this thing, AI, is actually helpful. And when it gives you something back, don't just accept it, correct it. Tell it what you change and keep going until the output is something that you would actually use. And this is the whole trust-building part because you're watching it, you're reviewing it, and then you're teaching it, and every single correction, every iteration, makes it better.
And step three, once that feels easy, start stacking bigger tasks together and connect Claude to the tools you already use, whatever you use every day, your Gmail, Slack, calendar, etc. There's probably a way to connect Claude to it or to connect any sort of AI agent to it. You guys saw in that first example when Claude pulled my YouTube analytics, I didn't have to go in there and export the data or copy and paste anything or upload some spreadsheet. I just said to Claude, "Hey, go get this data from YouTube." And it just did it. And because I have that set up and all of my other connections set up to all the other tools that I use every day, AI can actually see what's going on in my business and everything from there gets easier and better. You just want to do it in a way where it's safe and you always stay in control. Claude asks before it touches anything and you decide what it's allowed to reach and you give it certain permissions. And once you're automating real work, this is where you start keeping score. So, set a goal, like maybe this week I want to save myself 3 hours or I want 100 new leads per month. And then, track where you're at right now before you start using AI, and then check the number after, and the next month, and the next month. Because if you didn't hit it, you can improve the system and make it better. And that's really the only way that you actually improve is if you were keeping the data.
So, that's the path. You talk to it, you hand it one real thing, and then you just stack. Those are the first steps to start using Agent QA. But this space is moving really fast, and every week AI takes more and more jobs. But it's also creating a ton of new opportunities. So, if you want to start mastering how you can use Agent QA, you can join my free school community. I've got a full free course in there around building your own AI operating system, which is a great place to start. The link for that is in the description. But anyways, that's going to do it for today. I appreciate you guys making it to the end of the video, and I'll see you on the next one. Thanks, guys.