Designers are safe. AI lost.

AI is not replacing skilled designers or developers. If anything, there will be more designers needed, but of a different kind. Also, here is Slobgate posting jumbled text and AI slop on their official Instagram. Some 60 billion brands just don't care. And XX be like, "You never brushed your teeth with pineapple the pretty fluffy cream? " Well, I haven't. But GPT

image 2 can now design better than 80% of pixel pushers, which means junior and mid-level designers will have to step up their game or look for another job. And that is actually great news because design has become a bit of a joke in the last decade. So, we'll talk about that, too. AI is dramatically shifting how designers work. But I did ask some of my larger corporate clients, "When do they think my job will be AI replaced? " And spoiler alert, they laughed very hard. We're going to break down what they said, why, and then at the end of the video, I will share with you my professional workflow using AI. It helps me deliver better products, and it's not generative, so it's not making any visuals or images. But it does something that used to take a long time. And now the value of my work is way higher. And then at the very end of the video, I'm going to go through most insightful of your comments from the previous one. Let's go. There is a whole rather big

group of clients or people who will be just fine using tools like GPT image 2 and similar to generate a design. It is a sign of a big market shift, but at the same time, a lot of it will remain exactly the same. Let me explain. If all

you do is move boxes around on a screen in Figma, then sorry to say that, but that part is actually the easiest to replace. And I think Figma has to step up their game really soon, and not necessarily go into coding, but into some other direction that will be actually beneficial for skilled people. Because if you do just that

I generated these two app UIs because I worked on both banking and pregnancy apps, so I have a base for comparison. They're strong seven out of 10 to eight out of 10. Heavily dribble inspired with some small mistakes, but overall pretty good. They even make some sense UX-wise as part of a flow. Now, non-designers will prompt get something like that and then build it into a product. Designers will take that, generate 30 more versions with completely different prompts, and then use their taste and their logic and their knowledge of design fundamentals to merge those 30-plus versions of it into something even better, and then re-prompt again. And that process is going to be likely more expensive than $20. But we'll get to what some of my bigger clients said about outsourcing that kind of work to AI. It's pretty hilarious. The main

mistake here is assuming design is assembly. So, putting boxes on top of boxes or inside boxes and adding a blue and purple gradient, and then design is done. Assembly tools have changed over the years. They've gone from very simple like Photoshop into dedicated component-based design systems [clears throat] and all that stuff with the modern design tools. And every time when there was a change, we got faster. We were able to do more in less time. The main benefit is now that we have assembly figured out. We have way more time and energy and budget to do more and to do

better. And yes, the client pays the same amount of money for a much broader output. But you know what? I always wanted to be able to actually do more, to have a project that makes sense on every possible level instead of skipping some part just to fit within the budget.

I recently had a couple of meetings with some of my larger clients, corporations employing at least a couple hundred people each. They are all using AI in their workflows. They use it as help with coding, with some marketing information, some copywriting, and even data analytics. I asked, naturally, "When do they think they will be able to use AI to do what we do for them at Square Block? " And [snorts] they all laughed because they said that even if AI becomes good enough to actually generate the level of quality and insight that we deliver, they would still prefer somebody else to use that AI because that way they are not responsible as a corporation for those outputs. We are as designers. Because they do have the expertise to evaluate these outputs in their main niche. So, if they are analyzing data based on their own patterns and protocols and methods, then yeah, they can be sure that they can evaluate and say when the data is wrong. It's simply their job to know these things. But they are not designers, and their own developers report that there is the industry average of 25% of horrible code or breaking the code when you prompt for too long. While they all like the speed and the flexibility of AI tools, they have a lot more at stake than small startups. They can't be responsible for releasing something half-baked or not fully developed, not fully thought out, or not fully evaluated. Even if it is AI generated, it has to be tested on multiple levels multiple times. So, it's not a small startup making a silly landing page and prompting for it. Stakes are much higher. But at the same time, we have seen high-stakes large corporations completely mess up and deliver AI generated full of hallucinations, reports, presentations, and documents. But that is not the norm. Most respectable companies actually can't afford to make mistakes. And you

know what? Those quick, cheap landing page generated people weren't even the main target for most designers anyway because previously, they tried to find the cheapest designer possible to adjust a $20 template for them. These are roughly the same people. Maybe the group got a little bit bigger, but serious companies, no. They're not going to be prompting AI to generate a complex system or even a marketing landing page if they are serious about the job. When AI gets good enough to be able to generate these things, they will be asking a designer, a skilled person, to make that for them. So, the more I look at what's going on, the more I feel like the need for designers is actually going to grow a bit. But mostly because

you can do so much more with these tools right now. And I'm going to share something really cool from my own workflow in a minute. All these tools are enabling us to go above and beyond. Still with the human insight, human feedback, human expertise, oversight, but you can now do just so much more

and it's pretty exciting. I recently made my own AI skill to do complex professional flow diagrams. They are great for three things: client collaboration at an early stage, having a very well-documented flow for the developers, and also exporting them later as MD files if you want to generate some parts of the wireframes as well. I'll show you a post factum diagram for Pitchformance because when we did this, the tools weren't as powerful yet. But just knowing how everything connects together and having the ability to tweak and change and modify it allows us now to build the platform even better. And by the way, we just added AB testing to it. So, now you can run AB tests on your own website, improve the conversion, and grow. A good diagram starts with precisely written user story. So, if you start by generating something with AI at this stage, you already lost. You're not a designer, you're just silly. I write mine from the user's perspective, but I add precise annotations that explain technically how this part of the user story is handled. So, I add notes about how the back end might work, how the data might be stored, what data is private, what is public, all and everything that is necessary for the AI to fully grasp and understand the flow to make as few mistakes as possible. And then, as you add the MD skill and you upload the user story, the magic happens. AI generates a very nice, zoomable, scrollable canvas with all the connections, and a lot of it is wrong. So, I made it in a way that's fully editable by me. And then I can just go back in, move stuff around, and rewrite everything. But still, a process that used to take about a week to do fully right, now takes about a day, maybe two. So, the main time saver here is actually me manually drawing rectangular boxes and arrows between them, which is a pretty stupid and boring job. I much prefer to have them already created and then analyze what's in them, modify, write about it, come up with new ideas and expand on them. The key thing is to not accept what AI made for you from the start because it works on averages and every design that I'm making is not supposed to be average. It's supposed to be the best possible design that I can make. So, I can start with an average scaffolding and then build something extraordinary around it. And I can do it because I know what I'm doing. I'm using AI consciously, not as a designer. I'm using it as a tool to outsource the boring parts like drawing rectangles and drawing little arrows. And that gives me more time to make things better, not faster. You can recreate a similar workflow if you want just based on what I've shared in this video. But if you want more, if you want the entire workflow, it's also coming to our Square Bot Blueprint community in the next update. So, stay tuned.

I believe that if you never change your point of view, you're not really that smart. I constantly look for other arguments or insightful ideas that help me build on top of what I already know and basically be better. So, let's go through three insightful comments from you guys from the previous video. And I'll try to do this as a part of every video because there is some serious gold in the comments now. If you have stuff to say, let's hear it.

Sometimes a cheeseburger from McDonald's is fine and sometimes you want to get some homemade delicious burger and sometimes you want a three-course Michelin star dinner. Yes, that is true. And in a way I touched on this in this video because the people who want to quickly and cheaply create things with AI were the same people who did that using templates and Fiverr. But it's all about the stakes. For many of the larger companies, the stakes are simply too high to outsource not even the thinking or not even the making, but being sure. Currently, generative AI creates the middle of the bell curve, but it's not exactly the middle. It's the middle shifted a little bit towards the quality part because these tools are being guided. Which is good news for people who want to create something on a budget, but at the same time, it means that the entire part of the average and average plus is being expanded to the sides now. So, there's going to be more things that are good enough, which means that they might not be good enough anymore, at least not to stand out. This is a great comment because this shows that if you are willing to learn and try to understand everything around what you're doing. So, if you're a designer and you want to understand marketing and understand coding, you can use AI to improve what you know about these things and even learn them a little bit. Which means it broadens your spectrum of possibilities. The problem is that most people don't use it that way. Most people use it in the laziest way possible to just generate stuff fast and cheap. And these people are the people that AI will actually make dumber and less productive in the end. And most of these people will not know if the result that they got is good or not. All right, like, subscribe and

let's talk down there in the comments. Have a beautiful day.