Gemini Canvas: Vibe Coding a Tetris Game

This one’s a quick demo of Gemini’s Canvas feature — Google’s AI tool that generates working, runnable code from natural language prompts. I wanted to see how far it could go with something non-trivial, so I asked it to build a playable Tetris game from scratch. No starter code, no templates, just a plain English request. It generated the full thing — game logic, rendering, controls, scoring — and produced a working preview in under two minutes.

The bigger picture

The Tetris game itself is fun, but the real takeaway is what this means for how we communicate ideas. If you’re pitching a concept to a client or trying to explain an interaction to a teammate, there’s a massive gap between a static mockup and a working prototype. Tools like Canvas close that gap almost instantly. You can describe what you’re imagining, get a functional version in front of someone, and have a much more productive conversation about what works and what doesn’t. They’ve also added a share button, so you can fire off a working demo to someone without any deployment or hosting.

Where it fits in the toolkit

I don’t think this replaces knowing how to code — understanding what’s happening under the hood still matters when things break or when you need something precise. But for rapid ideation, for getting a concept out of your head and into something tangible, this is genuinely useful. It’s especially powerful for designers who have the vision but maybe not the engineering chops to build a quick proof of concept. The barrier between “I have an idea” and “here, try this” keeps getting thinner, and that’s exciting.