It’s 1994. Jeff Bezos is in his garage, grinning like a mad scientist, clicking through a website that he created himself that sells books online.
His friends nod politely, say “cool idea,” and then go back to flipping through their glossy Barnes & Noble catalogs.
People knew about the internet. It was in the papers, on late-night talk shows, even a subplot on sitcoms. But almost nobody understood it. To most, it looked like a slow, expensive toy. The few who did get it were already busy bending it into something bigger, breaking stuff, shipping experiments, and quietly laying the groundwork for everything that came next.
Copy-Paste Future
AI feels exactly the same right now. Everyone knows about it. CEOs put “AI-powered” in their press releases. Half the internet is pasting ChatGPT outputs into docs, sometimes leaving the “as an AI language model” placeholder text in there.
That’s not understanding AI. That’s using it like a faster typewriter. A shinier calculator. A tool that makes existing processes a little more convenient without ever asking what new processes are possible.
The 1% Mode
There’s a small group of people treating AI the way Bezos treated that garage website. Not as a gadget, but as clay you can shape.
We’re testing new models as soon as they are released, sometimes breaking them, seeing what sticks.
Currently I am actively working on:
- ProductLogz - A feedback management tool for product teams. We are reimagining & trying to built a tool that’s relevant for teams in this AI era. More on this later.
- Rocket Journal - a journaling app that uses AI to help you write your thoughts and ideas.
- QuotesMatic - a ios mobile app that helps you find quotes for your social media posts.
Each project is a tiny (or not so tiny) experiment to see how far I can push AI. And I’m not just talking about building AI tools for others. I’m using AI all the way through to build these products myself. 🤓
I mean Cursor is open almost all day for me, writing code and wiring things up while I work on the high level architecture.
Lately, I’ve started deep diving into Claude Code too, trying to see how much faster I can ship if I let it handle more of the heavy lifting 😎
Honestly, Claude Code & Cursor helped me tackle coding challenges that once felt beyond my reach. With its intelligent assistance, I can transform ambitious ideas into working solutions, bridging the gap between what I envision and what I can actually build. It’s like having a skilled coding partner who helps me push past my limitations and achieve things I never thought possible with my programming skills.
I’m learning on the go, testing what sticks, and experimenting with areas I barely understood a few months ago. Even the terminal doesn’t scare me anymore. Warp has made it feel less like hostile and more like a co-op partner that actually helps instead of throwing out cryptic errors.
I am not sure if I am the 1% or not. But I am sure I am building something that is relevant for the 1% and the future
The difference is simple. Most people are asking AI for answers. The 1 percent are asking AI to build with them. Testing every model the minute it drops. Running multiple agents side by side. Building tools on top of tools. Overall trying to ride the AI wave. 🌊
The New Frontier
1994 was not about sending emails faster. It was about reshaping how we buy, sell, talk, and share. Most people did not see it until years later.
2025 feels like that again. AI is soft clay. The rules of building are getting rewritten almost every day. We can bend reality a little, see what happens, and push it further.
Everyone knows about AI. Almost nobody understands AI.
We are the 1 percent again. Might as well keep shipping like it.
p.s: this post was inspired by Danny Aziz tweet.
