Top AI Product

Terence Tao wrote Java applets in 1999 to teach complex analysis and linear algebra. Web standards killed them years ago — dead code, unrunnable, for over a decade. On July 11 he blogged that an AI coding agent ported roughly two dozen of them to modern JavaScript in a matter of hours. The post hit 230 points on Hacker News within a day.

The scoreboard: one bug introduced, two bugs found

Here’s the detail that matters. The agent introduced exactly one minor bug (a drag-event glitch) — and found two bugs in Tao’s original 1999 code that he never knew existed. His verdict: a net wash on code quality. Some applets came back better, like a Besicovitch set visualization upgraded from monochrome to color.

Why this is the perfect use case

Legacy migration is the task nobody wants: tedious, low-glory, low-stakes if something breaks. Tao’s point is sharp — for secondary visual aids, the acceptable bug risk is low enough that agent-written code just works. He even shipped apps he’d abandoned in 1999 as too laborious, including a special relativity visualizer.

This is Tao’s second big endorsement of AI-assisted math after Lean formalization. When the strongest mathematician alive treats coding agents as routine infrastructure, that’s the signal.

You Might Also Like

• Cloudrouter Gives Your ai Coding Agent its own Cloud Machine and Thats a big Deal
• Vercel Agent Browser Might be the Smartest way to let ai Actually use the web
• Vet by Imbue the Safety net Your ai Coding Agent Desperately Needs
• P0 Bepurple ai Finally an ai Coding Agent That Actually Ships Whole Features
• Agent Kernel Gives any ai Coding Agent Persistent Memory With Just Three Markdown Files