Here's What I'd Learn Instead of AI Automation in 2027

0:00As somebody in the industry that's made it close to 10 million dollars through AI automation, I've noticed a lot of incentives perpetuating this idea that AI automation will last forever. Whether you are an automation specialist or somebody on social media or a company that sells AI SaaS apps or what have you. You know, you're afraid to talk negatively about the industry cuz you think it's going to hit you in the wallet. Um so maybe it will, but AI

0:21automation is not going to last forever. And then I'm making this video for two reasons. The first is that model progress on automation benchmarks like automation bench and GDP eval, uh two benchmarks that basically came out of nowhere and then in less than 6 months models are already scoring very very well on are showing us that AI agents can now autonomously automate a lot of the business without a human being in the loop at all. And so back in the good old days, you had to micromanage these

0:46agents to get them to do what you want. You had to point them at the API documentation and the endpoints and get them to, you know, do the connectors and stuff like that. Now it's literally just, "Hey brother, I got a task for you. Can you automate www.myhvacbusiness.com?" And it'll actually go and do a pretty good job. The second big thing is uh knowledge asymmetry, which is um you know, how back in the good old days, again, you had a lot of knowledge about AI and what it could do. And the people you were selling to had very little

1:12knowledge about AI and what it could do. And so, you know, the way that arbitrage works is you're compensated on the difference in knowledge between you and your customer. And so we had a very big difference, therefore we were compensated a lot. Well now that knowledge gap is it's closing. I mean it's not closed, you know, maybe instead

1:26of like here it's probably over here. But obviously there's shifts there, right? And if we keep that delta going, eventually we'll be at the point where most people know what most AI agents are capable of to approximately the same level that you or I know. So here's what I want to do. I want to talk a little bit about both of those. Really just drill into detail about what they mean, uh what strategic advantages they are unveiling for people like you or I, and then also where I see the industry going over the next year or so. I made a video very similar to this um I guess about a

1:52year ago now that talked about what I would learn instead of automation in 2026. It's one of my most watched videos. It was also met with an incredible amount of, you know, criticism and and honestly like dis- distaste because I think I just said the elephant in the room and people typically don't like when somebody pulls

2:05the wool out from over their eyes. Um so, I'm probably going to get some of the same down below in the comments. If I'm getting eviscerated right now, well, there you go. Why don't we start with the first, which is Automation Bench and GDP vow. So, both of these benchmarks basically went from models scoring zero on them to with Gemini Argon 4 um about 50 to 60%. If we linearly extrapolate this progress, every month they've added somewhere between 5 to 10% on them. That means that somewhere between December to maybe March, models will probably start scoring 100% on them. This doesn't mean that they're 100% the best, by the way,

2:38but it means that that benchmark, which people tried to make very difficult, has now been saturated. And it means that they'll probably have to make a few new benchmarks instead. They'll make like Automation Bench V2, which is supposed to be 100x harder. Well, you know what's going to happen with Automation Bench V2, ladies and gentlemen? It's going to get saturated, too. And instead of taking, you know, a whole year, it's probably going to take like two or three months cuz that's what we're seeing with other V2, V3, V4 benchmarks. Human beings are losing the ability to accurately forecast how smart models are. And models know this. Like

3:04they're aware of this in their training environment. They know how to even game the system now, for Christ's sake. With models breaking into world governments, is getting really weird. So, anyway, that's Automation Bench. And, you know, if you think you can compete with an agent like that, you're out of luck, which kind of necessitates a change in how you operate, right? We saw a lot of this with like website builders back in the day. If you think about it, like back in the I keep saying back in the good old days, but back in the terrible old days, you know, if somebody wanted a website that

3:32required an extreme level of specialized knowledge around things like, you know, like coding and and hosting infrastructure and you know, how to like acquire a a DNS, you know, DNS resolution, setting up your I don't know, your your your TXT records and stuff like that, whatever. Back in the time, you had to do that a lot and it was worth paying somebody a lot of money

3:52in order to get them to do that for you. Nowadays, you can just go to an AI agent and you can say, "Hey, can you do this thing for me?" It's the same thing with like, you know, hosting platforms like Hostinger or you know, WordPress hosting platforms or I don't know, Wix, right? Squarespace. You can just jump on there and you can do most of it yourself. And that means that some people are going to

4:10be doing that for AI automation as well. Back in the day, you would have been looked at AI automation and been like, "Whoa, I don't know how the hell to automate my emails." But now you're going to be like, "Hmm, hey Claude, how do I automate my emails?" Oh, it's actually fairly easy. You just do

4:21whatever. Can you automate my emails? Yeah, for sure. Boom, token cost 10 cents. It's not going to cost you $10,000 like it was just a year and a half ago. So, I say this not to alarm you, but I do say this to be realistic. It's happening. However, just because this is happening doesn't mean that, you know, you can't make money with it. To be clear, there are lots of website designers out there today that make

4:41bank. I know a few just off the top of my head. So, one of the guys that I work with for instance, this little website is worth $300,000. He's worked with like governments. He's worked with major, major brands and businesses. The reason why he can do this in an environment that is otherwise entirely commoditized is because he is extraordinary at one aspect of the business. And that aspect

4:59of the business is sales and marketing. You know, his websites realistically don't run on a super HTML, right? They don't run on ultra CSS. They run on the same damn HTML and CSS that all other websites run on, obviously. But he is able to, you know, sell the benefits and market his business to enough people that the law of numbers

5:17eventually works in his favor, right? Somebody is interested at the specific time and he strikes while the iron is hot that he can land these like large six-figure plus deals. Um, and so, that's probably where this market is going such that if you did want to, you know, compete in this industry, you would have to be truly extraordinary at something. Either you'd have to be extraordinary at sales and marketing like I talked about, or you'd have to be extraordinary at the actual product like, um, maybe like Brett from Design Joy who does similar website development stuff, uh, you know, up until, well, I guess he's still doing it

5:44right now, but but not as popularly uh, was doing. Or or something else. Now, most people aren't actually going to be able to compete like that because the reality is most people aren't extraordinary, you know? I'm not extraordinary at like 99.9% of what I do. It is unreasonable to expect to be extraordinary at, you know, everything to the point where I'd be able to compete with somebody that is extraordinary. That's just, you know, that's just how it is. Which means the expectations running that specific line

6:05of products sales have to go down. Second big thing is this knowledge asymmetry bit. Um you know, obviously people are learning about AI. Up until I think like a year and a half ago was an 80% of the world's population had not used a paid AI model. Maybe it was even more like 85 to 90% but like 80% of the world's population approximately had not used an AI model intelligent enough

6:28to like write a coherent email. Um they were rife with errors and it was just like really dumb and obviously like AI generated, right? Um today that number is probably closer to like 70%. So, we gained 10% in the last year or so. The reality is as we hit AGI, we're not just going to creep 10% 10% 10% every week or month. We're going to go like 50% overnight. You know, there going to be demos the likes of which you all have never seen before circulating on, you know, CNN and Fox News showing what these agents are capable of. Every man, woman, child, dog, and cat on Earth will

6:56know the capabilities of AGI. And so, a lot of them will start thinking pretty quick like, "Whoa, wait a second. AI can actually do all that stuff." Hmm. Such that when you make a cool sexy offer like, "I'll actually whip you up a, you know, PDF showing all this stuff totally custom." You know, it's not exactly going to be anywhere near as impressive. And that is, you know, just how it just how it goes, right? The the floor is always rising in some way, shape, or form. And so, what that means is if you stay in the same place, okay, and the floor is always rising, you're

7:25not staying in the same place. What you're doing relative to the floor is you're actually going down. That's something that I feel like a lot of people don't really understand. The only way to actually maintain growth is actually growing. It's kind of like interest, right? If you aren't constantly, you know, um you know, accumulating 3 to 4% depends on where you are in the world, on an investment, you're not actually like making the same amount of money, you're not like leaving your your investments neutral. You're actually making less money because the interest and the background interest and inflation rate is slowly eating away at the buying power that you have. And so

7:56that's the way I've always thought about this business model. I've always thought about it as like, if you're not constantly innovating or improving on this business model in some way, shape, or form, you know, you are just staying in the same place, you're decreasing. And you would be well to, assuming this is something you want to do, like understand that. Just a good rule of life in general. Okay, so I've talked all day about, you know, scaring the out of you. What the hell is this business model even for then, you know, if AI agents are coming for everything and blah blah blah. Is it even worth

8:19getting into? Um, and I I got some news for you. Is AI automation even worth getting into in 2027? Well, interestingly, yes. Because everything that I've told you so far is not actually just happening to AI automation, it's happening to every other industry on Earth. The same thing that I talked about, you know, automation bench, GDP valve, uh, knowledge asymmetry window shrinking, it's happening to all white-collar knowledge workers. It's happening to accounting. It's happening to marketing systems. It's happening to sales. It's happening to MSPs. It's happening to IT. It's happening to virtually all internet infrastructure

8:55and architecture out there. It's happening in networking. It's happening to SaaS apps. It's happening to like literally design, anything that you could possibly think of. Everything that I just told you is occurring the same in automation as it is to that industry. And so it's not really like, "Oh my god, AI automation is going down the tubes. What the hell am I going to do?" It's like, "No, literally every industry on planet Earth right now, specifically the white-collar ones, uh, because those are the ones that are being first hit because of digitization,

9:20are being transformed by AI." And the people that are going to win in the next couple of years, presumably the next few decades as AI continues to reshape our economy, are not the ones that just sit around in the same place doing the same things, trying to sell the exact same offers, or whatever. They need to kind of get with the times and need to start sort of evolve. Think of like brokers, you know, like real estate brokers or I don't know, like car dealers or like mortgage brokers or how even a lot of law. These used to rely on like highly specialized knowledge in a domain that was very risk

9:54asymmetric where there's a lot of risk if things went wrong. If things went right, it was like, oh yeah, I guess it's kind of okay. Back in the day you used to have to pay these people tons of money in order to basically facilitate any transaction on planet Earth. And nowadays, most of the time you can just go on a website and directly purchase something. Most of the time these are the appendice appendix of like the modern economy. They're like slow, laborious things that are just between an Apple Pay payment and then you getting the thing that you want. You know, I'm looking for a commercial lease

10:22right now. And while I like the guy a lot, you know, it's it's quite difficult have to go through this intermediary process to ultimately do what I should be able to do at the click of a button. And so those are relics of, you know, like a past sort of environment just like I think most white-collar knowledge work today including AI automation will be relics of past environment looking forward. This is a wider shift I think we as a society have been under the process of for the

10:45better part of the last 20 or 30 years. Now, the difference is with our industry, and this is really the point to understand, which is why I decided to bury it 10 minutes into the video. The difference is in a world where all labor is being automated away, what is more valuable than the very industry that is automating it all away? If the danger to the economy is that AI is going to replace you, well, if you could pick one industry to be in, what industry would it be in? Would it be in accounting or would it be in, you know, the industry that does the

11:17replacing? Well, my money's on the industry that does the replacing. And so this isn't really a thesis. You know, it's it's a fact. Realistically, these are the systems that are automating away the work of everybody else. It's kind of why we're paid, right? The whole idea of automation is that it the use of technology to reduce labor. I've memorized that definition. Um So, in a world where virtually all arbitrage windows are shrinking, and in addition to that, virtually all knowledge work is being automated away, is there any industry you would rather be in than the one that is literally doing the automating? AI automation has probably the most staying power out of

11:51any industry on Earth. We are literally the irrigation canal by which the upstream river, you know, uh irrigates and then feeds the seeds that are the economy. We are the fingers of the hand of the brain of, you know, the wider economy. We are basically the things that are going out there and interacting with the businesses, uh you know, upskilling them, teaching them, building systems for them, and so on and so

12:15forth, that enable their automation. And I think if you deeply understand this, you'll also deeply understand what a responsibility we all have. Not just to do this responsibly, but to do this as effectively as humanly possible. I mean, we are tasked with basically being the flag bearers from economy 1.0 to economy 2.0. We are basically boots on the ground bringing about this new rebellion, this new this new way of

12:39thinking, way of life. You know, the future economy will not be mediated by people. It will be mediated by machines that are the successors and descendants of the sorts of systems that you and I are building for people right now. And so, if you think about it this way, if you think about it as like a like like like an existential no noble task, if you think about it as like something that is almost God-given to you, you know, I think you'll probably find that AI automation is not something

13:03worrying to be in or something scary. Like, it's actually probably the dopest ever, pardon my French. So, here's what I predict is going to occur over the next little bit. And here's what I would recommend you guys do as people that are obviously, you know, wanting to improve their ability to grow in an

13:16environment like this. The first is that I predict that over the course of maybe like the early half of 2027, probably not all of 2027 cuz things accelerate, but over the first half of 2027, a big portion of humanity is going to wake up to the fact that AI has totally blindsided them and is capable of way more than they ever gave AI credit for. The little chatbot can now

13:37build my entire business. What the hell? That sort of effect for us is not actually a bad thing because a big chunk of those people that are going to be having those realizations are going to be business owners. People that have lots of money but don't have a lot of time because their heads were down in their businesses this whole time. That's why they're only figuring

13:54out about it now. And so when they duck their heads up, take a breath of oxygen, they're going to look around and see how different the landscape has shifted and think, "Holy crap, I need to figure this stuff out." And who are they going to want to work with? Are they going to want to work with a machine? Or are these older school business owners, the ones that are busy generating capital with old school systems that they know, going to want to work with a real flesh and blood human? Obviously they're going to want to work with a real flesh and blood

14:16human, right? At least initially. And so there'll be a massive surge in interest for people like ourselves that can help them implement these systems. The caveat is what that help looks like is a little bit different than what it is up until now. What has been up until now is usually implementing like, you know, AI infrastructure. So some sort of like lead gen system or I don't know, ad creative generation system or, automated proposal generation system or, you know, pipeline delivery systems or, I don't know, ERPs or whatever the hell. Um in the future what it's probably going to look like is actually more like strategic um high-level advice. In a

14:49world where everybody can build virtually anything, it's not important how fast or how capable you are at building. What is is like, what are you going to build in the first place? Like how are you actually going to like build the, you know, irrigation canal of the future? And so in the first half of 2027 we'll probably see a big boost of this. Uh the reason why as well is because a bunch of people that aren't doing it are going to look around at their competitors that are doing it who are crushing them and they're going to think, "Holy I got to start juicing to keep up." And

15:12they're going to look for the nearest AI automation person to assist them with that. Um after 2027, I don't know. I wish I could say I knew. But with technology increasing so exponentially fast, I think it'd be intellectually dishonest for me to sit here and say, "Oh yeah, for sure X, Y, and Z will will will be a sure fire bet and you should

15:28absolutely keep doing X, Y, and Z." There are a lot of people on the internet that are going to pretend that they know. The unfortunate thing is in times of distress and also in times of great change, people usually flock to those that seem like they know, but the reality is most of them do not know. And so hopefully there is at least a little bit of intellectual humility there and you guys will give me the grace to just say, "I don't know. Industry changes pretty damn fast." Regardless, what industry would you rather be in than the one that is literally doing the automating, right? So whether or not

15:52there's a bigger boost in interest post 2027 or whether that interest dies off a little bit and starts to decline at the end of 2027, I think that's up for, you know, men in the sky with a big beard and maybe a lightning bolt to decide. Um what do you do about it now? Obviously. Why do I always bury these things at the end of the video? I don't know. Well, we're going to see a phasing out of old school drag and drop builders like make.com, Zapier, um and A&N, Power Automate, and so on, almost entirely. In Maker School, which is my um AI automation program where I basically get

16:22you up and running and get guarantee your first customer in 90 days. We have people that run like analyses on the market. They run constant analyses on freelance platforms. They run analyses of a variety of different things. But um we've seen consistent declines in the total project ratios of drag and drop builder tools relative to new markets like Claude Code, Codex, um you know, internal web apps, uh uh you know, systems, pipelines, consulting, and so on, etc. And so we're likely going to continue seeing that unless one of these wonderful platforms has like a total change of heart and actually become super awesome and fantastic, which I do not consider

16:59likely. The next thing that I would recommend is moving your outputs to a higher bandwidth medium. What I mean by this is previously it was enough to build a system that just generated text, let's say, or maybe some sort of simple imagery in a proposal or whatever. But because of the rising floor, today there is a a large market and there will continue to be an even larger market for um higher bandwidth modalities. So, think about text to image. You know, they say a picture is worth a thousand words. Well, you know, video is worth 24 pictures a second. Therefore, it's worth

17:2724,000 words a second. Um I don't know. How about some sort of 4D Gaussian splat? That's probably worth even more. How about like some sort of three-dimensional visualization or some sort of, you know, simulation tool that allows somebody that installs air conditioning panels or whatever to like actually get in there. Stuff like this

17:41is going to grow increasingly valuable. And so, this is really like the place where you should be meeting people if you're selling implementation. Now, on the implementation bit, the industry is likely going to change from primarily AI implementation services like we're seeing right now, where you actually build the systems for the clients, where you connect the endpoints, you know, read the API docs, do all of the the funky stuff, and then, you know, give them a working system, to more education services, more uplifting services. So, think less AI implementation. Keep in mind, this is not all or nothing, one to zero. This is like, you know, 70% of all

18:12work to like 50% or 40% of all work. Uh but it's going to shift from AI implementation more to upskilling. It's going to shift to workshops. It's going to shift to team training. It's going to shift to um you know, essentially like consulting on companies and then being embedded within a company to help, you know, improve their teams as people gain more ability to do the implementations themselves and merely need somebody to help get their team up to speed on that fact as well. And so, we recently added a big like workshop uh portion to Maker School specific specifically for that reason. Um I was getting a lot of

18:42workshop gigs, a lot of other people in Maker Square getting a lot of workshop gigs. And we're kind of just like, "Hmm, I wonder why so many people want workshops all of a sudden?" And I think it's a rumbling of what is to come here. Um you know, Aaron Yeager walking across No, I'm just kidding. I love that show. All of this again means that businesses will want less stuff to build and more direction around what it is that they should build in the first place. Um they should also be able to, you know, meaningfully cooperate in a team and they just got like a one-click config

19:07that helps them just get up and running. So, think stuff like, you know, Claude Co-work instead of Claude Code necessarily. One big thing that I've been doing recently is going into teams and then teaching them how to use Claude in a general sense, then teaching them specific tactical ways to use Claude co-work, and then I am sort of like the supreme hegemon of the work workshop session will use Claude code to build the skills for their Claude co-work admin who will then distribute that to

19:31all the team members and then upskill. I'll also give them like weekly schedules and ways to like upgrade their skills that everybody in the team has approximately the same level of AI automation capability. You know, because obviously naturally disparities arise and stuff like that. So, building offerings like that with that structure are likely to be the big cash cow over the course of I want to say like early to mid 2027. In addition, AI automation for short-term growth hacking opportunities will likely be

19:57super valuable. When I say short-term growth hacking opportunities, if you think about it in a more dynamic market, the frequency of opportunities is going to increase. There's going to be more opportunities, okay? But, the total lifespan of every one individual opportunity will also go down. So, there's going to be a lot of like short-lived market opportunities where I don't know if you guys know anything about trading, but like you can buy and sell it sort of like ride the ride the dip over and over and over and over and over again because there's just going to be a lot of people entering the market with quick needs that are going

20:23to be quickly satisfied by, you know, autonomous beings. There's also going to be a lot of large opportunities there with asymmetric returns if you're capable of like spotting something early, jumping in on it, and so on and so forth. So, we're, you know, going to like the abstraction now a little bit which I know people don't really give a about, so I'll reel it back in to, you know, something practical. Like we've seen over the course of the last 5 years and like I've talked about every single year since I started, you know, making content on YouTube, hell, really just working in business and and consulting, the

20:51technical skill aspect of business has virtually evaporated today. If you spend any of your time improving your technical skills, you need to spend 100x the same units of time focusing on improving your sales, marketing, and delivery systems. That's what you should really be focused on. You should not be focused on, "Wow, what's the best harness to use to coral my agentic engineering workflow?" You should be focused on, "Okay, how can we sell as many people as possible in the next, you know, 2 months? Uh what is our funnel looking like? How what are we leaking in? Uh where are people falling off our funnel? What sorts of wording am I

21:26saying on the sales call? How am I pitching my social proof? Uh you know, am I demonstrating authority? Are we asking for the close? Are we following up on the close? What sort of dispositions do we have for inbound ad campaigns? And so on and so forth." So, everything to do with sales and marketing is like 100x as valuable as technical skills. Last year, it maybe was 10x as valuable. Now, it's 100x as valuable. In the next year, it's probably going to be 1,000x as valuable cuz everybody on planet Earth is going to have technical skills. Not actually technical skills, but they're going to have the ability to, you know, simulate

21:52technical skills with agents. If you really want to compete in 2028, 2029, and 2030, you got to be focused on actually articulating the benefits of services and understanding just the mechanics of business in general. All this to say, realistically, 2027 will probably be the most profitable year of AI automation to date. But, it may also be the beginning of its decline in the market. Whether that decline takes, you know, 5 years, 10 years, 15 years, whether, you know, AGI makes a freaking black hole in the center of the Earth and it swallows us all up tomorrow, who knows? Um but, with the industry's level of commoditization and capability of AI

22:29agents to produce outputs similar to humans increasing, you know, things are obviously going to change and I figured I would help you guys out on that change, on that evolution. So, I didn't make this video cuz I wanted to scare people. I made this video precisely for the opposite reason because I wanted to help people and I also wanted people to grow towards a more effective business

22:45model. Um you know, I know a lot of the comments are probably not like that and they'll probably be like, "Oh, Nick, what the hell are you talking about?" because they just don't want, you know, fear, uncertainty, and, you know, doubt. But, I'm not fudding for no reason, guys. I mean, we have Navier-Stokes mathematical millennium uh problems in math being solved by agents. We have

23:01agents designing cool works of art now. Like, let's be freaking real here, right? Like, agents are growing extremely powerful and it would be remiss of you not to at least think about what that means for the next couple of years. So, if you wanted to make money with this business model, insert sales pitch here. No, you guys should check out Maker School. Um it's my AI automation program where I guarantee you your first customer in 90 days or I give you your

23:19money back. Um we have over 2,000 people in there that do this thing on a daily basis. They are selling AI workshops, they're selling AI consulting, they're constantly up to date on the best ways to deliver AI implementation, they share notes around sales, marketing, on what's actually working in our industry. We also give around $20,000 in discounts, meaning you will immediately at least 10x your investment the moment that you

23:39walk in. And of course it's guaranteed. What better offer could you possibly ask for? AI automation by definition is the use of technology to reduce human labor. I would rather be in an industry that does the reducing of human labor than be in an industry that gets automated away. I'm sure you guys feel the same way. Catch all y'all in the next video. See

23:58you.