A 7-year-old is building a video game with AI: what really matters when you build software without knowing how to code

🧒 A seven-year-old is building his own video game.

He doesn’t write a single line of code. He describes the games he likes to the AI, looks at what it produces and tells it: “you got it wrong, do it better”. The video has passed 800,000 views.

“Coding is a solved problem”: the sentence from the creator of Claude Code

In those same days, the creator of Claude Code, one of the most widely used tools in the world for coding with AI, wrote a sentence that sparked a big debate: “coding is a solved problem”.

Since November 2025 he hasn’t written code by hand, but he keeps building software every day.

It looks like the end of a profession. For people who don’t code, though, it’s the beginning of an opportunity.

What remains open: software engineering

AI has become very good at writing code. But building a product that works takes much more: understanding what is really needed, recognizing when the result looks correct but isn’t, getting real people to use it and figuring out what to improve.

The creator of Claude Code says it himself: coding is one thing, software engineering is another.

And that part, by his own admission, is still unsolved. AI-generated code keeps having errors, inefficiencies and needlessly complicated solutions.

The sentence and the distinction come from the 19 February 2026 episode of Lenny Rachitsky’s podcast, where Boris Cherny, the creator of Claude Code, explains he hasn’t manually edited a line of code since November 2025; on 21 August 2026 he reiterated on X that “coding is solved, bugs are not yet solved”.

Translated for non-developers: AI has made a skill that used to require years of study far more accessible. But it hasn’t replaced the judgment needed to decide what to build, how it should work and when the result is actually correct.

Where judgment comes from

That judgment comes from knowing a trade, a customer or a market.

It’s the advantage of a freelancer, a business owner, or anyone who lives with a concrete problem every day.

Four things to focus on if you don’t code

💡 If you don’t code and want to build something with AI, focus on what really determines the outcome:

1. Start from the problem, not the tool.

Define precisely what the product has to do and for whom, before you even open an AI.

2. Write the important rules first.

Pricing, data handling, legal constraints and critical decisions cannot be left to the model’s interpretation.

3. Test the product the way your customer would.

If something seems to work but you haven’t truly verified it, the job isn’t done.

4. Think about distribution.

A product nobody uses remains an exercise. Who will find it? Why should they try it? And why should they come back?

The barrier has dropped, judgment is worth more

The technical barrier that kept many people out has dropped enormously.

Now it matters even more to know which problem to solve, for whom, and how to tell whether you have solved it well.

What about you — the last thing you built or would like to build with AI: did you start from the problem or from the tool? 👇🏻