Let’s talk about my first real project — something I built from absolute scratch. I’ve had a lot of ideas. Most of them died before they even touched the keyboard. Some were too ambitious, some too time-consuming, and others… just irrelevant to the world we live in now. Every time the negative thoughts came, the projects went straight into the trash.
At first, my focus was money. I thought, 'Let’s build something that gets views, some ad revenue, a few clicks on AdSense.' But that’s where it hit me — the problem with such projects is that they don’t bring users back. You can’t grow something that doesn’t matter to people. And maintaining a website, even a small one, costs money — domains, hosting, AI integrations, all those fancy toys. AI is cool, but not when it burns a hole in your wallet.
So, I had to find an idea that was feasible, meaningful, and ideally… cheap.
I’ve always wanted to know what’s happening around me — what people are talking about, what the country is obsessed with — without having to go through a dozen sites. I don’t have that kind of time or patience. So I started working on an app that gives trending news from around the world in one place.
Imagine this: you open the app, pick a country, and instantly see what’s trending there — news generated based on the top searches. Simple, clean, global.
The foundation for all this is Google Trends. It’s a tool by Google that shows what people are searching for on the internet. You can even filter by country to see what’s hot right now. So, I thought, why not take that data and let AI turn it into readable, bite-sized news content? That’s where it all began.
Now, I wasn’t ready to throw money into something that might not even survive a week. I wanted to build with max effort and zero cost. Not the best business plan, but hey — that’s how most experiments start.
I started researching how to build a website. ChatGPT became my personal guide. It suggested Next.js as a framework. Since I already had some experience with JavaScript, I decided to give it a shot. What surprised me was how easy website development has become. Server-side rendering? Handled. API integration? Smooth.
And then came AI — the ultimate crutch. I found myself using it for everything, from bug fixing to writing small chunks of code. With Mac’s native integration for ChatGPT, it became way too convenient. Honestly, I’ve become lazy… but efficiently lazy.
I went with Next.js for the frontend and Vercel for hosting. Their free tier was perfect for my small experiment. I launched the app — no custom domain, but zero cost. That’s what I call balance.
Now comes the fun (and painful) part — integrating AI to generate news. Initially, I tried scraping data from websites. Sounds cool, right? Well, it wasn’t. Every site has a different layout, structure, and logic. I was dealing with hundreds of countries and multiple trending topics. It was chaos. Eventually, I gave up scraping and turned to AI again.
But even AI wasn’t cheap. I had 1300 countries. Let’s say 20 topics per country — that’s over a thousand AI calls. Updating weekly would cost me roughly €32. Not ideal. So I cut down the list to 100 countries — about ₹300 per week. Still not free, and that bugged me.
Then I had a better idea: generate the AI news only when a user actually clicks on a country. The app would save the result in the database, so the next time, no need to call AI again. Smart, right?
Well, kind of. The downside? The first load takes around 1.5 minutes. But that’s a problem for another day. For now, I’m fine with it — because spending money on something uncertain? Not my vibe.
This is just Part 1 — the concept, the idea, and the chaos before implementation. In the next part, I’ll dive into how I actually implemented it, what went wrong, what worked, and how I’m planning to evolve it further.
Because at the end of the day, this isn’t just about building an app. It’s about building a legacy — something that speaks, something that lasts.
And maybe, just maybe, this is where Phantom Pyli starts making his mark.
