Now That Claude Does Everything, Here’s What AI Can’t Replace

The scariest thing about AI is not knowing what it will replace and what will still be valuable in the future. So, in this video, I'm going to walk you through the four biggest shifts towards value that AI can't replace and give you tactical ways to capitalize on each one. And no, this is not an AI doomer video. It is for people who want to understand where value is moving and get a step-by-step playbook, whether you are a business owner or an employee. And if you don't know me, I'm Austin. I'm not just a content creator. I ran an engineering team at J.P. Morgan to replace Palantir. I was a COO of a tech startup with over $25 million, and now I work directly with top AI brands like Anthropic. Now, before we get to the first shift, we need to establish a formula for quantifying the value provide. So, value is equal to what you get divided by what it costs you, time, effort, and money. And with AI, we have to ask ourselves a question. Are we providing more or less value than if someone just had AI do it? That question reveals what AI can replace, what it can't, and where your value needs to move. And throughout this video, we're going to revisit this equation. Shift number one is the decision premium. The hard part of your job used to be doing the actual thing. Now, it's knowing which thing to actually do. To showcase this, I work with a sports media brand on their app with over 100,000 users, and I helped them manage a team of four engineers. The hard part isn't actually building the product. The hard part is deciding what to build and how to build it. There's a premium on decision making. I'm not getting paid to write the code, I'm getting paid to help inform the product direction and how these features get built. A quote from Anthropic CEO Dario exemplifies this. >> Do you automate 90% of the job? Great, people are 10 times more productive in the other 10% cuz they're 10 times more leveraged. But, eventually it gets close to 100%. Now, the sequel to that is, well, then you have to find something else for them to do. >> So, the doing gets absorbed and the deciding expands to fill your whole day because that's what matters. And it's not just about making decisions, it's about making great decisions. Sure, AI can put hundreds of variations in front of you, but you can't put hundreds of options in front of your boss, your customers, or your audience. For that same sports brand, we recently redesigned the whole website, and when we launched it, customers were immediately happy. They said, "This feels more like a brand and less generic AI." So, you only have so many opportunities to put things in front of an audience, so you have to make sure you're making the right decisions. And as the speed of building accelerates, good decisions will compound, and bad decisions will quickly decay. This is what I call the decision premium, and the decisions that you make have never been more important because of this general product velocity. So, you have to become an expert decision-maker because that's what's valuable. So, how can we use AI to help us make better decisions? First, you want to understand the tradeoffs and the costs of being wrong. Decisions should be analyzed based on the impact of getting something wrong. So, Jeff Bezos calls it one- or two-way doors. If it's reversible, which is a two-way door, move fast and don't overthink of it. If it's a one-way door, that means you can't reverse the decision, then think critically before you complete the task. Once you understand this categorization, the most effective thing you can do is just ask yourself, "Is this a one- or two-way door?" and that'll impact how much energy you spend on that specific decision. The second is you want to use AI to time travel. The hard part about making a decision is that you never actually know if it's the right decision until after the fact. So, let's use AI to time travel in the future to improve the quality of your decisions. The best way to do this is you want to create a skill that I've actually spoken a lot about on this channel, which is called your internal focus group. This focus group should be designed around the exact person that this decision impacts. To do this, here is a prompt that'll walk you through exactly how it works. So, let's say you're creating an end-of-month report for your boss. You would have AI interview you to establish who your boss is, what they like and don't like, and then you will take that information to create an AI clone of your manager to provide you feedback whenever you call the internal focus group skill. The same pattern can be used for really anything, but the key here is picking the right person to be part of the focus group. So, this is usually whoever is on the receiving end of whatever decision you're making. This could be a boss, a customer, a client, a teammate, etc. The third way to make better decisions is be a level two decision-maker. This is my favorite one because I tell this to everyone I work with. I always say that there are three levels of decision making. Level one is where you just do things that you're told to do. So, let's say you were told to design an email campaign for an upcoming product launch. You would have the task and you would just complete it. This is no different than if a person took whatever directions they gave you and told AI to do it. And this type of thinking leads to average outputs, which quoting directly from Naval, a famous entrepreneur, there is no demand for average. Level one decision making has no value because AI can just do it faster and cheaper. Level two decision makers ask the next question and think a bit deeper about the problem without having analysis paralysis. To outline this for the email campaign example, they'd understand who the audience is and think about different segments. Then they would elect to make one email for one segment and then another email for a different segment. They thought deeper about the decision and made a better output. And this is the sweet spot. This is the Goldilocks zone. You're not overthinking, but you're not underthinking. And then level three decision makers, this is the over analysis, hyper fixated on details that don't actually matter or provide any value to a business. This is the type of person that spends 45 minutes debating if they should start the email with hey or hi. And I know everyone watching this has a person in mind when I say that. The people who are able to become level two decision makers will be significantly rewarded in the AI age. Now, before we get to the second shift, which is all about capitalizing on an arbitrage opportunity. A key thread throughout this whole video is knowing when to use AI and when not to use it, which brings us to today's video sponsor, Anthropic, the team behind Claude, which I am absolutely hyped to be working with. They just released a new beta feature that I had to do a double take on. It's called reflect and it's so interesting because it does the opposite of what every other app does. See, every app you see is designed to maximize how much you use it, but Anthropic built this to help you better understand and manage your usage patterns. It does this by first highlighting your own usage outside any of your incognito chats, your health conversations, or source files. So, it shows you what you actually work on with Claude and how that shifted over the past 1, 3, 6, even 12 months. It looks at that usage and asks its own hard questions back to you. For example, what's one thing you want to keep doing yourself? And if you've watched my channel, you know that I talk about a concept called intellectual obesity, which is where people aren't using their brain enough because AI is just doing everything for them. Now, this is a unique opportunity where Anthropic has decided to try and help you use your brain more, not less. That's the first way it helps you manage usage. But, the second is it has a feature called quiet hours and break reminders. So, you can set the boundaries on your own usage and Claude will remind you about them as you use the tool, which is a feature I desperately need to help me get better REM sleep so I'm not using these tools all night. And the third way is they built it with digital wellness experts from MIT Media Lab, Boston Children's Hospital, and the Family Online Safety Institute. So, the product itself feels thoughtful in terms of putting the human first. You'll find Reflect under Settings in Claude on web or their desktop app. It's available for free, Pro, and Max users as long as you've got memory turned on. So, go try it by clicking the first link in the description. Shift number two is the arbitrage window. There is a massive gap right now between what people actually use AI for and what is actually possible with AI. And to show this, here is actually a recent text from my buddy saying how he was blown away with how easy it was to vibe code a website. He previously wasn't aware and now he's aware. And this knowledge gap is what I call the arbitrage window. With any new technology, there is the rate of adoption for the general public, and then there's the actual rate of the advancement of the technology. And the reality is with the technology, specifically in the AI age, it advances quicker than people can learn or adapt to it. And so, what this creates is a massive arbitrage window where you're able to use AI to complete tasks almost instantly. But, for people who are unaware of those capabilities, it provides a ton of value. So, to spot opportunities where you can provide value that may seem easy to you, but are valuable to others, use this prompt. This is a way to step out of the bubble and ask you the question of, "What do you know how to do with AI today that 2 years ago would have blown you away? That's the reality, right? If you're watching this video, you are in a bubble. So, 2 years ago, what would you think is amazing? Now, if you're wondering, on screen, you'll see a bunch of industries and where I see the biggest arbitrage window in each specific category. The goal is to find a thing that is trivial for you to do, but valuable for one specific person, and then do it proactively. So, let's say you're employed, right? You could build an internal tool for your manager that maybe he didn't ask for, but you do it proactively because it solves a problem that they had. Or, if you're running a business, you could offer your clients free value that is easy for you, but may blow them away in terms of what they were expecting. And this strategy is something that I do with most of my clients. If their website looks a bit outdated, I'll just revamp it essentially for free. It provides them a ton of value, and it's pretty easy for me to do because I know how to use AI. And here's an objection I get every single time I bring this up. If AI can do it, why is it valuable for me to do it? So, going back to the value equation, value equals what you get divided by what it costs. Run the equation from their perspective. If they don't know that AI can do this very easily, then they just get the outcome, right? What they get, and it's essentially free or doesn't cost them anything. So, obviously, it's going to be valuable. Who cares how it got done, or who cares how easy it was for you to get it done? AI is the tool, and the output is what matters. But the key with all this, and none of this is for short-term windows. There's a reason that this is so important. It's not for the short-term gains. It's about creating an angle that lets you establish a long-term relationship with the right type of people. I'll cover why this is so important later in this number three is optimize your private loops. Everyone has access to the same AI models, so how can you differentiate yourself to provide unique value? Microsoft CEO put out a memo that describes this perfectly. And if it sounds confusing, don't worry, I'll unpack exactly how this impacts you right after. Every company is going to have to build human capital and token capital. Human capital comprises the knowledge, judgment, and pattern recognition of its people. Well, token capital is the firm's AI capability it builds and owns. Importantly, human capital does not become less valuable as token capital grows. It only becomes more valuable. I believe human agency will be the driver of token capital growth. Without human direction, you have compute running in circles. This means the real opportunity is not in picking the best model, but instead in building a learning loop on top of models where human capital and token capital compound. You can offload a task or even a job, you can never offload your learning. The future of the firm is the ability to compound the learning across people and AI. Okay, that was a lot, a ton of fire bars in there. And there's another fire bar that I didn't mention that I'm going to reference in a little bit, but simply put, the business that wins in the AI age will have humans who best allocate tokens or AI intelligence to help the business grow. And there's one line that I want to call everyone's attention to. And it's about the opportunity of building a learning loop on top of models where human capital and token capital compound. A learning loop is the process of taking the industry specific data that you know and have access to and applying it to AI tools. This sounds complicated and I didn't fully understand this until I recently experienced it. One of my clients is an engineering firm that generates a report in New York City called Local Law 97. We built a cloud skill called Local Law 97 Report Generator and the first 10 to 20 runs, it had issues with different edge cases. But each time that we iterated and we provided specific feedback that only this engineering firm had, were we able to create a skill that repeatedly generated the report exactly how we wanted for all of their clients. This was our process of merging our human capital engineering expertise to evaluate the results with the token capital and AI intelligence to create the report. And we created a skill that is the intellectual IP or the intellectual bridge that this business can use going forward. The value of this is obvious, right? Revisiting that value equation. What you get, an engineering vetted report, and what it costs is essentially free. So this is valuable for everyone involved, the employee who created it, the business that now has it, and the customers that now receive the final product. This process of iterating and improving your AI system based on the private data is what I call optimizing your private loops. So, how do you actually do this? First, find the thing that you do over and over. You need feedback cycles to improve the system. A task just happens once, then there's really no juice for you to squeeze. The second is you want to turn this into a skill, not a prompt. Anything that you do repeatedly needs to be a Claude skill. This is not negotiable. Now, I've made videos on my channel on how to do this that I'll link at the end of the video, but here is a prompt where I'll look at what you've done in the past, and I'll identify potential skills that you can then create. The third step is repeat and capture learnings. Each time you run a skill, it's an opportunity to improve the skill. This is the most important part of this whole shift. That's the data that changes the final result. So, here's a prompt that will help you improve your skills as you use them based on feedback you provide in a conversation with AI. The other three shifts that I cover, I would suggest doing them ASAP, but you can do when you want. This one can't. Each recurring loop that you don't use is is a missed opportunity. It's just the reality. Now, before we get to the next section, I want to bring in another quote from the Microsoft CEO on the importance of this private loop. This loop becomes the new IP of the firm. Every improved workflow generates better training signal, which accelerates the accumulation of knowledge unique to the firm. The companies that build this early will have an advantage that is hard to replicate regardless of any new individual model capability. Sheesh. Bars from the CEO. Absolute bars. This gets me hyped. Now, before I get to the last shift, if this is your first video of mine, welcome to the channel, but if this is your second or more, you know the drill. This is our anti-slap agreement. The visuals, the testing, the hours of research that go on in this video, it's entirely built for humans, not for AI token scrapers. So, all that I ask is you subscribe as part of this agreement to help this content reach more people. Also, every video I give away a Claude Max subscription, so this video's winner is barefoot bond 82 for building a gamified personal finance experience. Comment below with whatever you're building, a recent feature you built, or any problems you've run into. Every video you comment on counts as its own entry. And also, side note to all this, I am looking to hire someone to work directly with me to help me scale my business. So, the link to apply to that is below. Now, back to it. Shift number four is the most important one, and it builds on the previous three. Shift four, proof of work. Everything we just covered is about providing more value, but this section is about convincing someone that the value created is worth it. In the sea of AI slop, how do you differentiate yourself as someone who can confidently solve the problem that the person is looking to solve? To illustrate this, let's look at the job market for a second. Recruiters are drowning in perfectly polished AI-written resumes. This is to the point where a lot of them say they genuinely cannot tell a strong candidate from someone who just used AI to help them with their resume. So, instead of reading what lands in the inbox, more of them are going out and sourcing people directly and leaning on things like actual portfolios, GitHub projects, paid trial periods, even video generated. This shows an accelerated trend of something I've been preaching for about 3 years since I first started creating AI content. Proof of work is 100 times better and more valuable than any sort of credential. And proof of work comes in different forms, and I've had broken it down into four tiers, where each tier is harder to do but more effective. The key here is that this applies to both employees and business owners. You constantly have to provide value, no matter what industry you're in. So, tier number one is showcase. This is you telling them what you can do. It's a resume, a portfolio, case study. To stand out, you want to zig while others zag. So, when I interview candidates, don't send me a resume. Send me a cloud artifact as a way of showing your proficiency. Tier number two is demonstrate. Show the value directly, not describing it. So, say you're applying for a job, prove you can do the task with a real example. Let's say you're selling a product, show the exact outcome that they want. Tier three is a human vouching for you. So, this is someone else's recommendation, right? We all know how valuable this is. This is harder to fake, and it's only getting more valuable as people want to trust people, not AI robots. And tier number four is personal lived experience. This is the most important, and it is also the simplest. Has that person already gotten value themselves directly from you? So early in this video, I stressed the importance of the arbitrage window. This is the primary reason why. It's not so that you can get a short-term win. It's so that you can start building real relationships with real people proving that you can provide value. This is the type of thing that compounds over time and that's the exact reason that I spend close to 60 hours across my entire team on every video I produce. I want to provide you with real value for free so that I can grow my channel. So use things like that arbitrage window as a easy way to capitalize and provide free value to the people around you. So with these tiers in mind, I put together this prompt that will help you identify unique ways to showcase your value to the people around you. Whether you want to or not, you have to be able to sell yourself in the AI age and these are the best ways to do it. All right, let's recap the four clear shifts going on. There's a decision premium. There's no demand for average. You have to become an expert decision maker and that will become significantly more valuable over time. You have the arbitrage window. There will be a time where what is easy for you with AI isn't easy for others and this lets you capitalize in the short-term positioning yourself for long-term success. Private feedback loops. As said by the Microsoft CEO, people who successfully use private feedback loops to optimize token capital will create a competitive advantage that will be around for the foreseeable future. Proof of work. Anything you do now is questioned because of AI. The best people understand the four tiers of showcasing what value they can provide. Now, if you got this far, you may be wondering how to optimize the Claude skills that you create and so for that, go check out this video where I break down exactly how Anthropic's team creates Claude skills and how you can follow their exact process step-by-step. Whether you're technical or not, that video is for you. I'll see you over there. Peace.