"Two milestones shipped, 63 plans across 11 phases, 25k lines of code. Gradinka is in good shape."
25 hours. I sit on my station and scratch my head in disbelief. It feels like a win, an adrenaline rush. The fucking clanker spat a full serverless system for 25 hours! My Claude Code Max x5 subscription is far from being depleted, and I spent a weekend and a few hours throughout the next few days to get this done.
Now, you might wonder why I am so ecstatic. Right, everyone, these days, talks about vibe coding, people sharing their "magical" tricks to squeeze context, keep it lean and aim for the goldilocks of having good output.
But this is different
For years, I've worked with many different developers. Some - brilliant. Others - mediocre. I have deep connections with some of them, and with a few of them I work till today. One thing that I was always struggling with was that for more than a decade and half, despite countless hours working with Flash, PHP, HTML, CSS, JS/TS, Databases, Servers, Hostings, CPanel, DNSes, Firewalls, APIs, Web3, Cybersecurity and whatnot I was always telling myself: "Bro, you are better as a TPM / Manager / Client Relations. Stick with that. Code for pleasure, or to put a fire away." So I did not personally recognise myself as a "coder/developer", simply because I was doing 100 things.
I just wanted cool projects we can cut through and excel. I always kept my name, giving my team the confidence to work with me and providing the business with a single point of contact. And businesses recognised this - building trust and just delivering no matter what was my mantra. Small company with a hardcore team. **** So many nights I was there patching something after someone, covering a gap, tying loose ends...
What is a developer?
Going from battle to battle, from challenge to challenge, I slowly started to notice something odd. I was not a developer, but I had to tell developers more and more things on how to do. Yeah, I was coming with a decade of experience, which is quite a lot of coding under my belt, and I'd just bumped into all kinds of obscure situations in my projects, but it was still odd for me that there was something obvious a "real developer" couldn't see.
At the same time, I started to have specific projects and needs that required professional JS/TS developers and skills. At some point, I decided (Angular 1.4) that it would be really cool to start coding my own things outside the already mastered WordPress custom themes. Around the same time, I learned that reading the documentation can be tedious, but for a newbie learning new technologies, it is a must.
I wanted some specific things coded on specific technology. And everyone I stumbled across tried to convince me to do it otherwise. Practical example: I wanted a hybrid mobile app coded on Vue. By that time, all freelancers I spent weeks with (note project was well described) wanted tens of thousands of dollars to make it, and that's not so bad, but:
- "I don't have experience with Vue" ****
- "Let's do it in React Next, React, Flutter"
- "Capacitor / Cordova / Phonegap will never work"
- ... and ton of other bullshit.
Desperado to Badass
At that point, I simply understood: if you come from this background and want things your way or the highway, then you need to do these yourself. So I started to learn in depth Vue3. I couldn't, and I can't to this day, chew on React. I understand the code, I know the logic and the way it works, I know Next framework and I don't want to have ANYTHING in common with them. **** Simply don't like how React embeds the templates. Angular - off the game after 1.4.
Fast forward 2021-2026 I have more than 10 projects on Vue, and Nuxt. Built nationwide game for FMCG client with serverless MongoDB Atlas, have coded interactive role playing games with turns and complex logic and yeah - made my hybrid mobile app with Capacitor and Nuxt - looks like Duolingo with Rive and Lottie animations GSAP, local SQLite and ton of other miraculous technology.
So, necessity made me a Fullstack Developer with more guns and experience than quite a lot of other people. And the best thing - I could do, evolve and make miracles in this ecosystem, especially Nuxt, that untied my hands serverside and provides so many optimisations.
Struggle
While I was super excited and really hungry to become a badass Fullstack developer, I was struggling with all those tasks, as I continued to push my agency forward (and we have a lot of corporate clients, looking for WordPress websites), making side quests (like 1 year fantastic challenge producing video content for EU course with Prezi Presentations, After Effects and merging this into interesting and captivating 12+ hours course.)
I was happy and grateful, but the only light I saw back then was around 2024, when I started adopting AI heavily into my workflow. It was not optimal, but it was speed x5 to x10. Yeah, old training data, loops towards Nuxt3 and Vue options API, and shit like this, but it was muuch faster. Still didn't had so much of advance, untill...
Tipping point
With my experience growing up and adopting more tools, refining Cursor and Windsurf folders, crafting Guardrails and Cline boundaries, it seemed that the technology capped. I am talking 2025 and in this year I had some great and awesome challenges with VPS infrastructure, automation tools like Flowise, n8n, Active Pieces, and still pushing interactive systems. Fast forward to November - I came across Claude Code, and this changed everything. It was so fascinating to see the way it works:
- Agents
- Skills
- Commands
- MCPs
- Orchestrations
- Safe areas
- Autonomous work
I really sensed that this was the turning moment. I started experimenting with Fullstack Nuxt applications by setting up increasingly complex Claude Code frameworks throughout November. And then one of my friends and colleagues, an extremely clever CEO we have discussed with these new opportunities, told me: "Just give the model trust and tell precisely what to do, no need for excessive semi-automations via Node scripts, yml declarations, frontmatters and complex documents".
I enthusiastically bought a Claude Code MAX x20 account, and all of these above couldn't tap it, couldn't even load it. I broke down the process of creation to a simple Feature Request / Tasks / Revisions / Sign Off process with some very clear and standardized templates, and was able to spin off some pretty complex stuff like complicated functionalities. The best part - the latest framework I created was able to make super clean and documented code, solve and refactor old code.
I had to write and review quite a lot of FRs and tasks to confirm that the things work fine, but the productivity was so high, and the errors - well, they might took me some time to address myself or iterate, but compared to standard use of Cursor and Windsurf, this was incredibly faster and more stable.
GSD
After pushing this framework around and flexing my task-definition skills, I came across GSD (Get Shit Done), and the rest is... history. The way this is built exceeds my framework in complexity, but the process is so smoothly designed that it is a breeze to build something you have in your head, you know the technology, and you have placed a solid foundation as a stack.
My head exploded. After some adjustments to my specific needs, I built a project that had been waiting since 2008! A social platform for the green city areas with a complex map, user management, sponsors (CSR), ratings, stats, audit trails, and map polygon drawing. Yes, of course, there were some pretty nasty bugs, a couple of outdated libraries, and incompatibilities, but man... 25 hours, lying on my sofa, calmly explaining what's needed, how it's needed, how it shall be planned, what shall be considered... Security audits and hardening, tests, all in place.
Exciting and Frightening
I am so excited right now, I really am. I have maybe 10 different projects I put on shelf, because I did not have time, because it takes too much goddamn time to lay down the foundation simple and redundant code... and now this is gone. My hands are untied, and I can leverage this technology to the fullest, mastering my own stack and tools, building projects for communities and businesses so fast that I get chills when I think about it. No more redundant developer spaghetti code to decipher; now I say what and how to have it.
I am frightened too, as I realise this will kill an industry, lead to more layoffs, and create more security holes if used by inexperienced developers and juniors without training. It will spin off a plethora of problems, not to mention rogue AI and Cybersecurity. Someone can optimise his/her Pwnagotchi, develop hideous micro tools for Flipper Zero, or craft a dark cyber weapon (have no doubt - you can develop 3-4 harmless parts and then merge into something vicious). And while I would not do such a thing, I see a clear and huge dark side.
How things will evolve. I really can't foresee how things move so fast. But one thing I know for sure - we will adapt, we will have to adapt to survive. I might not work in tech in 10 years. I might be a techno freak helping NGOs. We will see, but this is the actual state of AI engineering right now - 20-50x faster than a couple of years ago.