Dave Plummer has returned to his most famous creation — but this time, the code was written as much by a machine as by him. The new Task Manager, available now as a beta for Windows, macOS, and Linux, started life as a 107-page specification that Plummer fed to Fable, an AI coding platform. After a three-and-a-half-hour first pass, the first iteration of the application was produced.
"I picked Task Manager as a simpler target to see how ready AI is for primetime… or if it would just degrade to slop," Plummer told The Register. "And it turns out it depends on what you ask for, largely! And what you test and instrument. It's like having a very ambitious and precocious intern, but like the classic genie in the bottle, be careful what you wish for."
The new version goes far beyond the original's bare-bones process-killing functionality. It offers deep system insight similar to today's Windows Task Manager, but with more metrics and drilling capabilities. Visually, it evokes 1990s media player skins — Plummer deliberately styled it after his Pioneer RT-909 reel-to-reel tape deck. Users can switch fonts, themes, and even adjust a bloom-and-saturation slider that goes up to 11.
Rather than taking the common Electron route for cross-platform apps, Plummer opted for native UI on each platform: Swift on macOS, Direct2D/DirectComposition on Windows, and Qt on Linux. "It's more work," he said, "but worth it in a task manager or something that has to have game-like response times." The performance payoff is evident: the Windows version is roughly half the size of Microsoft's built-in Task Manager and uses half the RAM.
The code is currently proprietary, with a paid Pro version planned that will add more features. Plummer has not set a price yet, and there may be additional beta releases before the final launch.
While Plummer insists the project was never intended as a challenge to Microsoft, the new app's superiority in several areas — more features, better performance, higher customization — inevitably invites comparisons. For developers and power users, it demonstrates what can be achieved when AI is paired with clear requirements and domain expertise.