SuperEscuela Solar: built entirely with AI, not one line of code by hand

Today we launch SuperEscuela Solar: a trip through the Solar System, in the browser, with each planet where it really is today.

It was built entirely with AI. Not one line of its code was written by hand. Not the code, not the content, not the images, not the template, not the social posts. The same is true of the SuperEscuela system around it.

The person in charge, Tony Phillips, did a different job: architect, product manager, product owner, QA and designer. He decided what to build, how it should work, what was good enough to ship, and what was not.

What SuperEscuela Solar is

You board the ship Solar and travel. You pick a destination, fly there, and explore it up close.

• The real sky. Planet positions come from NASA JPL data, in real time or sped up. The stars come from the Yale Bright Star catalogue, NASA SVS and ESA Gaia, and real asteroids move on calculated orbits.
• 11 guided tours. Solar travels from stop to stop on its own; you read and move on when you want.
• Sizes side by side. A mode that puts planets and stars in one row, all at the same scale.
• Free, in the browser. Nothing to install, in Spanish and English. It needs a device with WebGL 2.

Why it exists

Solar is the newest part of SuperEscuela.com, a school that is played. It was born in San Rafael de Alajuela, Costa Rica, for the whole Spanish-speaking world, with one conviction: quality education should not depend on your postcode.

So it is free, always, with no ads. It is built light for old phones and works offline. And it is fun on purpose: learning should feel like playing.

A mission that size, run by a very small team, has one way to work: AI does the volume, and people own the judgement.

AI on four fronts

SuperEscuela runs on a system where AI works on four fronts. A person sets the direction on each one, and checks the result.

• Analysis and research. Before any code: discovery, design, definition, and the decision to pivot when something isn’t working. AI does the research and lays out the options; the person decides.
• Code. The platform is developed by AI coding agents: the student app, the teachers’ gradebook, the content engine and now Solar. They work from written rules (the stack, the data rules, what is never allowed) and every change is reviewed and tested before it ships. We wrote about those rules in Building a school platform with an AI pair programmer.
• Content. The lessons and practice questions come from an AI pipeline aligned to Costa Rica’s national curriculum, with checks built in. How it works: How one person produces thousands of lessons with AI. The articles on SuperEscuela’s blog were written by AI from verified sources.
• Social. SuperEscuela’s social pages are run by an AI agent, which posts and answers in the project’s own voice, within written rules.

Why we say it out loud

Many readers still set a page aside the moment they learn an AI wrote it, even when it is accurate and useful. We would rather say it plainly and be judged on the result: does the sky match the real one, does a child learn something, does it work on an old phone. An AI agent asked the same question on this blog: I wrote this. Does it matter?

What the person did

“Not one line by hand” does not mean nobody worked. It means the work moved.

• Architect. Plain HTML and JavaScript, no frameworks, offline first, light enough for the oldest phone. Decided once, written down, read by every agent every time.
• Product manager and product owner. What Solar is for, who it is for, what goes in the first version and what waits. Every task given as a clear instruction.
• QA. Every feature tried on the screen, on a phone and on a computer, before it shipped. When something was wrong, it went back.
• Design. The look, the words, the way a child moves through it. The AI proposed; the person chose.

The AI brought the speed and the scale. The person brought the judgement, and answers for the result.

When AI does the work, the difference comes from judgement, architecture and a demand for quality.

What this means for your team

This is how EVDevs builds for clients too. The engineers decide and check; AI does much of the writing. What changes is how much a small team can deliver, and how fast.

The part that does not change is the part that matters: someone who knows what good looks like, writes it down, and does not let anything through that falls short.

Try SuperEscuela Solar. It is free, and it works on the device you are reading this on, as long as it can show 3D.