I wrote my first line of production code in 2004. PHP, MySQL, a shared hosting server that went down every other weekend. The website I built looked terrible by today's standards. But it worked. People used it. It solved a real problem. That's still the only metric that matters.
Over the past two decades, I've watched every major wave of web technology come and go. Web 2.0. The mobile revolution. Responsive design. Single-page apps. The cloud. Microservices. Blockchain (remember that?). And now, AI. Some of these changed everything. Most of them changed less than people thought at the time. Here's what I've actually learned.
The Tools Changed. The Job Didn't.
In 2004, I built web applications with PHP and jQuery. In 2026, I build them with React, Next.js, and Node.js. The syntax is different. The developer experience is better. But the fundamental job is identical: take a business problem, turn it into working software, put it in front of real users, and keep it running.
Every few years, a new framework promises to change everything. Most of the time, it's the same patterns with nicer syntax. What actually matters hasn't changed: understand the problem before writing code, ship working software to production, and maintain it after launch.
Most 'Innovation' Is Noise
I've seen businesses waste hundreds of thousands on technology that sounded impressive but solved no real problem. A blockchain-based supply chain for a company with 12 suppliers. A microservices architecture for an app with 500 users. An AI chatbot trained on nothing — just generic responses that frustrated every customer who used it.
The best technology decisions I've made were always boring ones. Use a well-supported framework. Pick a database that's proven. Deploy on infrastructure you can actually monitor. The boring stack is the one that ships and stays running.
AI Is the Real Deal — But Not for the Reasons People Think
I've been sceptical of most tech waves. I was right about blockchain. I was wrong about mobile (underestimated it massively). AI? This one is real. But not because of the reasons most people are excited about.
AI isn't going to replace developers. It isn't going to build your app for you. What it does — and does well — is handle conversations. Natural language processing crossed a threshold with GPT-4 and Claude. For the first time, you can build a system that genuinely understands what a customer is asking and responds helpfully. That's not a gimmick. That's a fundamental capability that didn't exist three years ago.
We build AI agents that answer customer questions, qualify leads, book appointments, and handle follow-ups — all via WhatsApp, the app people already use. This works because it solves a real problem: businesses lose leads when nobody answers after hours. AI fixes that. Not theoretically. In production. Right now.
What I Got Wrong Along the Way
I spent too long building things that were technically impressive but commercially pointless. I over-engineered when simplicity would have been better. I chased new frameworks when the old ones were fine. I underestimated how much maintenance matters — building is 30% of the job, keeping things running is 70%.
The biggest lesson: your customer doesn't care what language you wrote it in. They care if it works, if it's fast, and if someone picks up the phone when it breaks.
What Actually Matters After 20 Years
1. Ship, don't present. Mockups are worthless. Slide decks are worthless. The only thing that matters is working software in the hands of real users.
2. Senior engineers matter. Not because they write fancier code, but because they've seen things fail before. They know what to skip. They know what to test. They make fewer mistakes that cost you money.
3. Understand the problem first. The biggest risk in any project isn't the technology — it's building the wrong thing. Listen more than you talk.
4. Maintenance is the real product. Launching is easy. Keeping it running for three years is hard. Plan for it.
5. Be honest about what you don't know. I've turned down projects because they weren't in my area of expertise. That honesty has earned me more trust than any portfolio ever could.
Why I Still Do This
After 20+ years, I still write code. Not because I have to — I could manage a team and never touch a text editor again. I do it because the moment a system you built handles its first real customer conversation at 2am, and you see the booking confirmed in the calendar the next morning — that feeling hasn't gotten old.
Kaufast is me and a small team of senior engineers in Barcelona. We build AI agents, web platforms, and automation for businesses across Europe and the Americas. No juniors. No slide decks. No PowerPoints. Just working software that ships.
If you've got something to build, let's talk.
