tools.theuws.com

1984 — An 11-Year-Old and His Atari

In a small town in the Netherlands, an 11-year-old kid named Richard unwraps a birthday present that will change the trajectory of his life: an Atari 600XL home computer. It comes with a chunky keyboard, a cassette deck for storage, and a manual that promises you can make the machine do anything — if you learn BASIC.

The excitement is immediate and consuming. Typing commands into the blue screen and watching the computer respond feels like discovering a superpower. PRINT makes words appear. GOTO creates loops. SOUND makes the POKEY chip sing. Every line of code is a small act of creation, typed character by character on a keyboard that clicks with each keypress.

Over weeks of after-school sessions, Richard builds JIM — a cowboy quick-draw duel game. Two blocky figures face each other on screen. A countdown ticks. Whoever presses their key first wins the draw. The POKEY sound chip fires off satisfying bleeps for the gunshot and a low tone for the loser. The graphics are crude: stick figures in a landscape of solid-color blocks. But it does not matter. The game is his. Every single line of BASIC, typed by hand, debugged by hand, saved to a cassette tape with its unmistakable screech of data.

2026 — Enter Claude

Richard is no longer an 11-year-old kid in the Netherlands. He is a tech entrepreneur who has spent decades building companies, managing infrastructure, and shipping products. He has seen every era of computing from the inside. And now, in 2026, he has a new kind of development partner: Claude, an AI assistant built by Anthropic.

The idea starts as a nostalgic impulse. What if JIM could live again? Not as a museum piece, but as a real game — one that honors the original while exploring what modern technology makes possible.

Here is the part that still feels surreal: the entire game was built with three prompts. Three carefully written plan-mode prompts to Claude Code — each one describing a chunk of the vision in natural language, each one triggering a cascade of architecture decisions, file creation, and implementation. The first prompt laid out the dual-mode concept and produced the complete game engine: state machine, input handling, scoring, audio systems, the retro Canvas 2D renderer with CRT effects and POKEY-style chiptune, and the full PixiJS WebGL renderer with particle effects, camera systems, cinematic transitions, and a 10-level campaign with boss fights. The second prompt refined gameplay, added the boot sequence, and polished the modern mode. The third prompt integrated everything into this website — the hero section on the landing page, this story page you are reading right now, and the play page with its lovingly crafted CRT television and sleek modern monitor frame, complete with the little channel knobs and speaker grille — and deployed it live.

Three prompts. From first idea to a fully playable dual-mode game, hosted on a live website, with all the bells and whistles. No boilerplate grinding. No Stack Overflow rabbit holes. No three-week sprint to get a prototype running. Just a clear vision, described in plain language, and an AI partner that understood both the nostalgic soul of 1984 and the technical ambition of 2026.

(We did need a fourth prompt to add this very paragraph, because apparently three prompts was not quite enough to also brag about only needing three prompts. It has also been deployed live, naturally. Some things never change.)

This is not about AI replacing developers. It is about AI amplifying human creativity — enabling anyone to take an idea that has been living in their head for four decades and bring it to life. The 11-year-old who typed BASIC line by line would be amazed. The tech entrepreneur who described the vision to an AI and watched it materialize is just as thrilled.