Systems Thinking: Automate your entire job with Claude cowork

0:00Right now, everyone's talking about AI. There's a new tool launching every week, but the people who are making money with AI are not the people who are using the most tools. It's the people who understand systems. Because AI doesn't magically build businesses. What it allows you to do is automate processes. So, in this video, I'm going to show you the skill that makes AI actually powerful, systems thinking. And by the end of the video, you'll be able to take almost any business, break it into

0:26steps, and automate the entire thing. So, a lot of people think AI is replacing jobs, but what's actually happening is AI is replacing the tasks within jobs. Every job in the world is just a collection of tasks. For example, let's take a YouTuber. What do they actually do? They don't just make YouTube videos. They perform a series of tasks. They define their channel strategy, they brainstorm video ideas, they write a script, they shoot and edit the video, then they make a thumbnail and post. And each of those tasks require them to follow a structured process to complete it. So, for example, to create a thumbnail, you need to

1:02research winning thumbnails, you need to create a concept, you need to take a photo, and then you need to edit it in Photoshop. So, to complete the task, you follow a process. And once you start seeing work like this, everything becomes clear because AI is really good at completing structured processes. The problem is most people think like this.

1:21They say, "I want to start a business." But systems thinkers do things differently. They ask, "What processes create the outcomes?" And once you identify the process, you can design the system around it. Now, let me show you how to actually do this. So, today, I'm going to use the example of a podcast producer. Every big podcast in the world has a producer behind the scenes who's responsible for reaching out to guests, scheduling them, and doing research so they actually know what questions to ask them. So, we're going to automate this entire job end-to-end. The first thing you want to do is list out every single task that person has to complete to

1:55complete their job. So, that way the AI knows everything it needs to do in order. Now you can offload this to AI and ask you to come up with it yourself, but the more specific you can get on what you care about, the better it will perform. And so I like to map these out ahead of time, and this is the real skill in the AI world. It's your ability to think and map out a process step-by-step. And I'm going to show you how I would do that for a podcast producer right now. So I'm going to call this my podcast producer workflow. So the very first thing we need to define

2:22is our podcast strategy. What kind of guests and what topics do we cover? This makes it very clear on what my podcast is about. So is it an NBA podcast where I interview NBA athletes, or is it a business podcast where I interview entrepreneurs? So that's the very first thing we need to define is what is the podcast strategy? Next, I really need to understand what are the right type of guests to come on my podcast. I might not want to interview every single entrepreneur in the world. I need to get a little bit more specific. And so this workflow, I'm probably going to come up

2:51with an ideal guest profile who would qualify to come up with my podcast based off something like their social media following size, their business revenue, their age, or have they been on the podcast before. So I run a podcast and I interview 18-to-22-year-old kids who have made a million dollars online. That is my ideal guest profile. And so we need to define that in the AI system so the AI knows who to invite on the podcast. Now that we've actually built that system and we know who the best guests would be, so we've defined who the ideal guest is, now it's time for the AI to go out into the world and find

3:25all of the people who fit that criteria. We actually need them to do what we call as prospecting, aka making a list of all the people that we could reach out to that fit that criteria. So basically, we're going to have to choose the platform we want to reach out to these people on. That could be Instagram, Twitter, or email. Then we need to actually make a list of their guest name, the business model, their estimated revenue, and then their social media followers. This is all information

3:49based off step two. So there we go. Those are the first three steps. We came up with a podcast strategy, so we know what the goal of the podcast is, what the vibe is, who it's for. Then, we get the ideal guest profile, who would be a good guest, and then we actually do prospecting, so we come up with all the people that we could invite on the

4:05podcast. Now, what's the next step? Well, of course, you guessed it, we actually need to reach out to them and schedule them to get on the podcast. So, we're going to message them on X, we're going to personalize the message, we're going to invite them on the podcast, and offer a date to shoot. That is the goal of reaching out to them. So, once we've reached out to them, we actually need to schedule them on the calendar, so they actually know when to show up and where to show up. So, we're going to add a Google Meets link, we're going to choose a time, and then we're going to add the

4:31guest to the calendar. And then, last but not least, once they're scheduled, I need to actually create a research report for the host, so we can research their background, and actually propose questions that the host should ask the guest. So, this is the entire role of a podcast producer. He's going to come up with a podcast strategy, he's going to create an ideal guest profile, he's going to create a list of all the people who fit that profile, he's going to reach out to them on Twitter, he's going to schedule the actual podcast shoot time, and then he's going to provide a research report with recommended

5:01questions to ask. This is the system that a podcast producer follows to complete his job, okay? So, there are six tasks, each one of those tasks requires a structured process, and all of these together is a system. So, this is what I mean by systems thinking, and this is the most valuable skill you can learn in the era of AI. Because now that I have this, it's very, very easy to execute and automate with the actual AI tool. So, now, let me show you how that's done. Let's imagine I'm running a podcast producing agency, and I just signed a new client who's already ran a podcast before, but he wants me to come

5:35help him manage everything. So, all I'm going to do is say, "Hey, I want to create an automated cloud workflow to run my podcast producer agency. Please build the system for my new client." And I'm going to paste his YouTube channel in. So, this is the input that triggers the workflow. Anytime I get a new client, I'm just going to put their YouTube channel in, and then everything else should be automatic from there, okay? So, now all I have to do is just paste in that system that we built, and just let Claude do its work. So, I'm using Claude Co-work with the actual Claude Chrome extension to automate

6:06this. So, that means there's no APIs, no security vulnerabilities. I don't have to connect anything. It's just fully automatic taking control of my computer and automating everything end-to-end, including the outreach, which I'll show you here shortly. So, now I put that in, and the very first thing Claude is doing is actually building out five individual skills. These Claude skills are

6:25basically telling Claude how to behave. This is kind of like the new era of prompting, where it's actually just creating a really detailed prompt. So, instead of me having to write some crazy prompt every single time, it's actually going to create a custom skill that it will pull from every single time, and this is actually saved inside of Claude Co-work, and so it's like your AI brain, so you never have to tell it again. So,

6:45as you can see, it made all the skills. So, we have the podcast strategy skill, which is custom-made based off of the podcast that I gave it. Then we actually have the ideal guest profile, so guest scouting, so it knows who to reach out to. Then we have the actual scheduling skill, so it knows how to schedule it. Then we actually have the X outreach skill, so it knows how to approach reaching out to them, how to research and propose questions to the host. So, in a less than 3 minutes, those skills were automatically made and now saved inside of my AI system. So, that is why

7:15systems thinking is by far the most valuable AI skill right now. Since I front-loaded all that work and I mapped out my process and built the system, the AI already built all the skills for me, and now all I have to type in is run the podcast workflow, and it's going to execute it for me. So, now the AI is running. It's going out on my browser, and it's researching all of the people who could potentially be good guests on my podcast. And as you can see, it came back with 10 candidates who are scouted, scored, and ranked based off of my criteria for my podcast. The first three

7:46it came back with are Zach Yadegari, Danny Postma, and Liam Ottley, all three of which I would absolutely have on my podcast. Another few recommendations are Shane Copeland, the founder of Polymarket, which who I've actually tried to get on my podcast before. Peter Levels, someone who I also think is really, really cool. The next step is to reach out to them, and as you can see, the AI is actually taking control of my browser, navigating to Twitter, finding each guest, and sending them a personalized DM on my behalf inviting them on the podcast. This AI will even handle responses and scheduling to create the calendar event, which is

8:19[ __ ] nuts. So, as you can see, it's typing the message and it just sent the first message to Zach, and just like that, the AI is officially inviting people and doing my guest scheduling for me. So, now that we've sent the average message, let's imagine that Zach came back saying yes, he would like to be on the podcast. Well, the last part of the system is to actually do a deep research report that actually proposes questions so the host shows up to the interview prepared. And so, as you can see, it does that. It creates a very detailed research doc telling you everything about the guest, what he's been asked on

8:49other podcasts, and even gives you suggested questions to ask, which is extremely helpful for podcast hosts. So, that's the entire system, but as you could probably see on my screen, there's a little bit of back and forth, and that is important when you're setting these up for the very first time. When you're going through this for the first time, you want to make sure that strategy doc is perfect. You want to make sure the ideal guest profile and the guests they're recommending and prospecting are people you'd actually want to have on your podcast, and you want to make sure that your outreach messages aren't cringey and they're actually written in

9:19a way that you would speak to people. So, now that you've spent a little bit of time manually making sure everything is perfect, it's actually time to automate this so it can work 24/7 hands-free. And all you have to do is ask the Claude agent to automate this workflow so this can run without me, no approvals needed. And what it's going to do is create something called scheduled tasks. So basically, you can create a task that gets triggered every Tuesday at 10:00 a.m. or every Monday or every morning. And so you can choose these sequential times to actually trigger the agent even when you're not there. And so

9:51if we scroll down, we can see the five scheduled tasks that this created. So on Monday, we're going to do pipeline review at 9:00 a.m. and on Tuesday, the agent is going to do guest prospecting. So every Tuesday at 10:00 a.m., it's going to search for 10 to 15 candidates who score against the 30-point rubric and build a ranked pipeline. So every single Tuesday, this agent is automatically going to make my list of potential guests. And then on Wednesday, it's going to automatically reach out to the three top candidates on X. It's going to reply and it's actually going to schedule them to book a time to

10:21actually shoot the podcast. And voila, you have an end-to-end AI agent running 24/7 without you needing to be involved at all and you've automated an entire role in your company. And that's it. This is now an AI agent running 24/7 on autopilot filling an entire job inside of my company. So this is why systems thinking is so powerful because if you can actually map out these processes one by one, you can automate an entire business. So now that I have this system built, anytime I get another podcast producing client, all I have to do is say, "Hey, run the podcast producer workflow for my new client and share

10:59their YouTube link." And it will build out all of the skills for them, it will build out the scheduled task, and do the entire process end-to-end. So most people are just experimenting with AI tools, but the real opportunity is designing systems that produce outcomes like this. This is why investors are suddenly interested in AI agencies. Y Combinator recently said they're looking to fund AI-native agencies. These are agencies where the work is performed by AI workflows instead of large teams because when you automate processes like this, one person can run businesses that used to take entire companies. Once you build a system, you can run it repeatedly, which means you're no longer

11:38selling time. That means you're selling the output of the system. If you want to learn how to build businesses like this, I am posting a free workshop where I will walk you through AI workflows step-by-step. I'll show you how to design them, how to build them, and how to turn them into real online businesses. You can join that community in the link below. In my next video, I'm going to be showing you seven different AI businesses you can start using systems like this. So, make sure you're subscribed, and I'll see you in the next

12:04one.