Swift, AI, and the future of software development

I’m struggling. One day, it feels like it makes no sense to continue developing apps, frameworks, and tooling. On other days, I’m super excited about what my team and I have been doing all these years. AI has turned the whole industry upside down, and as far as I can see, it has hit all developers hard. Let me explain a little bit…

Big AI corps are trying to tell you that AGI is here and that people don’t have to write code manually, but honestly, this is complete bullshit and a big fucking lie. The current state of LLMs is not as advanced as they want you to believe. Of course, they can do amazing things, and despite their mistakes, they are extremely capable of doing what they are supposed to do: generate things based on existing patterns. Since their training data is enormous and they can sometimes process information better than humans, they are able to see the connections between the dots and solve problems that baffle human minds. But this is NOT artificial intelligence; it’s still just a tool that can mimic things. 🪄

We haven’t seen anything like this before, but now it is the standard.

Don’t get me wrong: I’m using AI and haven’t stopped yet. I also love it because I don’t have to do boring, repetitive work, and I can focus more on software architecture and design. That’s the fun part, for me at least. But there’s a catch. Since I’m not coding, there’s something missing. I loved writing code, struggling with syntax and language nitpicks, and that has also disappeared from my workflow. I really miss it, but I can imagine that most people don’t like writing code by hand, and they find this new “vibe-coding” approach extremely satisfying.

You can get shit done fast. Extremely fast. It’s like doom-scrolling. You just prompt, and dopamine instantly hits your brain when you see something succeed. It’s an addiction. I can literally feel it, and I don’t like it. Over the last few weeks, I’ve been thinking a lot about getting back to manual coding, but it’s freaking hard to go back to writing code by hand when there’s a tool that can sometimes write even better code than you can. I never thought I was a good coder, but I’m good at recognizing patterns and coming up with simple solutions to complex problems. AI can’t solve this yet, IMHO. 🤖

There are other things I like about AI, such as grammar checking, rephrasing, and other small, language-related tasks that make my life easier. I’m not a native English speaker, and I make mistakes all the time, but AI can help me catch and correct them. This is also a huge win for me, but there’s a catch. I’m not learning new things anymore. It feels like I’m getting dumber and dumber with every prompt, day by day. I love learning new things, but I’m not “deep learning” anymore. This is what’s causing frustration for me. I sometimes ask myself: should I just build slop and stop caring about quality, like the others do? I feel a bit lost and confused because of this.

Also, many people are losing their jobs “because of AI.” Many leaders are firing people because they ask questions about the product, and they see this as an obstacle instead of a chance to refine, rethink and plan. Fine. AI won’t argue with you, that’s for sure. It’s all in the algorithm. I attended Appdevcon this year and I loved the talk about human relations in the age of LLMs. It was one of the best ones at that conference. The things happening around us might hit your motivation and bring instability. They hit mine hard, that’s for sure. 👎

So what can I do about this? Well, first of all, I’m not planning to give up or stop using AI. I’m just trying to find a way to make it work for me. I think I need a good balance. I realized that I’ve been multitasking a lot more since I started prompting agents because I have to wait until a task is ready, and that splits my attention. I want to be more focused and do less multitasking. Agents are extremely good at pattern recognition and can help me with my work. I just need a way to architect everything and think using the exact same pattern everywhere across the whole system. (Hint: I already achieved this, and it’s more or less working…)

Should I stick to one language or not? PHP is dead, JS is dead, and all programming languages are dead. English is the new programming language. We’ll see, but I wouldn’t dig the graves just yet. Of course, the industry is evolving, and new, modern languages are coming out all the time. As far as I can see, most languages are here to stay, and agents will generate apps by utilizing already existing languages. This is what they are good at: the internet is full of existing patterns, both good and bad, and LLMs already know how to work with programming languages.

So which language should I stick to? Swift? Or should I abandon it and go with Rust? Well, Rust is way more popular, JS is also way more popular, and the same goes for a dozen other languages, but I do believe that Swift is special. Most developers would say, “Why don’t you pick the one that suits the job best? Why care about this?” Honestly, I don’t know, but my gut feeling is that Swift is the way to go, and I have no intention of abandoning it anytime soon. I still admire Apple for its long-term plans, and I know there are ups and downs, hardware and software quality issues, and so on, but IMHO, they still have one of the best developer experiences across all platforms. (Feel free to disagree with me; I don’t care.) 😛

But you’re not even developing traditional apps anymore? Yes. I haven’t made iOS, macOS, or any other Apple-platform apps over the last six years. I was trying to build backend apps, servers, and web apps, and figure out how to get the best out of server-side Swift development. It’s a niche. It sucks sometimes. Other times, it’s awesome. It’s a real roller coaster, but I still have a lot of respect for the language and the community. Honestly, the community is trying to do its best to make Swift the best language for server-side development, and I’m so proud of it. Vapor, Hummingbird, and all the other open-source projects are freaking amazing, and I’m a bit sad that we don’t have a foundation to support these folks. Hey, Apple decision-makers, could you please listen to me and do something about this? Many thanks. 🤑

So why do I care so much about server-side Swift? Because I’m a backend developer. I started as a PHP developer, and we created a CMS in 2009. I’m a bit obsessed with content management systems and cool backend apps. I want Swift to be the best choice for building servers and web apps. I even established a company for this purpose, and now that AI is here, we’re struggling like hell, but I’m not planning to give up… I’d like to teach others how to architect backend apps with Swift. I’d like to build new tools to make server-side Swift development easier. That’s my mission, but I’m also a bit tired, so I need to find a better balance in my life too.

I’m going to become a yoga instructor. I’m going to start the course in September 2026. Wish me luck! I see this situation as a good opportunity to become healthier and more productive. I’m trying to establish a better work-life balance, and I’m trying to embrace AI while treating it as a tool, not as a magic wand that will solve all of humanity’s problems. It will help a lot, but in its current form, it’s more like a great automation tool than anything else. I bet you that 90% of current AI features wouldn’t require AI at all. Anyway, I’m getting a bit far from the original topic… 🪷

I’m here to stay, even if the world burns. (The current heatwave, with temperatures above 40 degrees Celsius, is crazy.) I’m going to do my best to teach people how to build better software with Swift and AI. You can’t avoid LLMs, but I believe that over the last year, I figured out quite a lot about them, and I’ve developed a pretty solid understanding of how they work and how you should architect your backend apps to work well with AI agents. I’m also going to publish new articles and upgrade this blog very soon. I just want to finish my CMS first, or at least bring it to a state where it can be used by anyone.

Sorry that I disappeared for so long. I haven’t really had the time to write articles. I was focusing on building Binary Birds and learning a crazy amount of things along the way. I love writing articles, and I still love Swift, but I don’t like the way people are using AI or what they expect from it. People also tend to be short-sighted, and they don’t plan for the future. I’m the complete opposite, but I’m fine with that. The only thing I don’t like is how much money we burn on useless things while people focus only on big numbers, traction, money, and shit like that. By the way, I loved this podcast, in which Oli told the story of the monk with the Instagram Live:

I don’t care about numbers, but I do care about helping others learn. I love to explain complex problems in an easy way; that makes me happy. I love to teach and educate. Anyway, I’ve gotten a bit carried away again. I don’t want to become even more emotional. I just wanted to write this quick article to give my readers a brief update about what’s coming. Feather CMS will be my next “big” thing in my life, and I’m excited to share this journey with you. Thanks for helping me and reading my blog. I’m also extremely grateful to those of you who have purchased my book or told me anything kind in the past. I appreciate this a lot. 🙏

This article was 100% handwritten by me, but its grammar was corrected using an LLM.