When Hostinger released their Horizons AI developer tool in March, I was excited but skeptical. I like Hostinger and didn’t expect a pure money grab, but many recent “AI tools” overpromise and underdeliver. I assumed Horizons would land somewhere between decent and impressive.
Horizons is marketed for non-developers without a technical background to build apps and websites. I decided to try both and share results, strengths, limitations, and usefulness. First, a quick how-to.
How to use Hostinger Horizons
Go to the Horizons site and you’ll see a simple chat interface. There aren’t model choices or advanced options — just a chat window where you describe what you want. You can attach sample files or images for inspiration and Horizons begins building immediately.
That “take it and run” approach annoyed me at first because I wanted to map out my idea before it started coding. I even told it not to start, and it ignored me. It’s a minor gripe, but a startup wizard to help non-developers flesh out ideas before building would be useful.
Overview of my experiments
I ran two main experiments:
– A relatively complex time-keeping app with advanced features.
– A one-page website with interactive scroll effects (rebuilding my portfolio).
1) Time-keeping app with advanced features (live demo linked during test)
What went right
From the first iteration, aesthetics were strong. Even without guidance on colors or fonts, Horizons produced a modern, attractive UI — reminiscent of Hostinger’s dark theme. I fixed some initial bugs via prompts and got the app to a point where UI interactions seemed to work: fields filled, choices set, and invoices exported as PDFs.
What went wrong
The core feature—time tracking—didn’t work reliably. The timer showed seconds counting while open, but when the tab was left and revisited, the elapsed time was far less than actual time passed (I left it for about an hour; it showed nine minutes). Some data didn’t populate in the dashboard. I didn’t try to fix these bugs further because I wanted to test the website experiment.
Still, considering I spent around three hours, getting a prototype this far is impressive. Five years ago, a similar prototype would’ve taken a developer weeks. Also, you can export the code and manually fix bugs if you have dev skills.
2) One-page website with interactive scroll effects
For this test I rebuilt my portfolio to add scroll interactions. I initially tried to paste the source of the inspiration site but exceeded the character limit by about 63,000 characters. To work around it, I took screenshots of the source and fed them to Horizons in batches. With each batch, Horizons extended sections and scroll effects. Then I gave it a screenshot of my live site and asked it to extract the text and rebuild the site in the template style we’d been working on.
I had Horizons replace placeholders with real links and images and correct some transcription errors. In total, it took only 16 prompts to recreate the one-page site from scratch.
What went wrong
A few minor issues popped up, most were fixed with follow-up prompts. One small visual detail — two words with a strike-through effect — couldn’t be rendered correctly; after one failed fix attempt I asked Horizons to delete it. Everything else was temporary and resolved through prompting.
Is Horizons worth it?
Yes—Horizons is a game-changer in many ways and offers value across skill levels:
– Non-developers with no technical background can build simple apps and sites with relative ease. Expect to make adjustments, but the goal is reachable.
– Non-developers with some technical knowledge will go further; using browser dev tools helps when you hit walls.
– Developers may benefit the most. Prototypes that used to take weeks can be spun up in hours. The ability to export code and modify it makes Horizons especially useful. It would be even better if you could import existing code into Horizons, which I hope Hostinger adds.
Horizons offers capabilities that go beyond what you typically get from ChatGPT or Claude when asking for code. Be realistic about expectations, use good prompts, and be patient — you can build impressive projects.
Horizons pricing
Hostinger offers a free seven-day trial with no credit card required. The trial caps you at five prompts per day (35 prompts total). That’s limited, but recall I rebuilt my portfolio in 16 prompts and got an initial version of the time app in 18 prompts. If you approach the prompts strategically, you can accomplish a lot in the trial.
You can’t publish without a paid plan, but the Starter plan is affordable. If you produce a useful prototype during the trial, paying for at least one month to keep your work is often justifiable.
Have you tried Hostinger Horizons yet? What did you think? If you built something interesting, share it — I’d love to check it out.
