
I Built a YouTube Shorts Clipper for $0.65
Cheap AI models replace paid services. I built a $0.65 YouTube Shorts clipper that turns long videos into vertical, captioned Shorts.
AI is far cheaper than most people realize — and it's here to stay. Its best use case is programming, which means learning to leverage these tools is the skill that matters now. Building your own software, running on your machine, is suddenly trivial. But it's not magic: you still can't vibe your way to success.
$0.40 in tokens spent: watch DeepSeek V4 Pro building Adobe Lightroom clone

Cheap AI models replace paid services. I built a $0.65 YouTube Shorts clipper that turns long videos into vertical, captioned Shorts.

DeepSeek V4 Pro only takes text. I gave it vision through a small Flash model in GritCode, so the agent can see the UIs of the apps it builds.

Adobe Lightroom doesn't run on Linux, and the free alternatives are clunky. So I built my own photo editor with AI, for about $0.40 in tokens.

Cheap tokens, cheap laptops, and the end of the CS gatekeeping era.

Fedora ships codec-free out of the box. Here's the fix to get hardware-accelerated H.264 / H.265 decoding back on a fresh Fedora 43 or 44 install.

How to display large datasets in wxWidgets without lag using wxListCtrl in virtual mode. Complete working example with 100K+ rows.