A LEGO fan vibe-coded a brick-finding tool with Codex

If you’ve ever built something with LEGOs, you know the sound: the clatter of those little plastic bricks as you hunt for just the right piece to finish your creation.

Boy Genius Report contributor and WordPress developer James Bruce must know that sound well. A self-reported LEGO fanatic, Bruce has described his passion for scooping up “old bargain bulk boxes” of LEGOs to “figure out what sets were mixed in there.”

What Bruce doesn’t love, he writes, is pinpointing the precise LEGO piece he needs when “there are 5kg of bricks to wade through.” If only there were a web-based tool that would help him track the bricks included in the vintage LEGO sets he picks up…

Yep, you can see where this is going. Bruce launched Codex, OpenAI’s agentic coding agent, and began firing off some prompts.

“Within 15 minutes, it had built a working prototype,” Bruce wrote. “Less than an hour later, I’d brought a domain name and uploaded the first version. It was rough, and didn’t work for larger sets yet, but the proof of concept was there, and it was already useful.”

A few days and several new features later, brickbacklog.com was up and running.  

“Despite two decades of experience as a web developer and programmer, it would have taken me several months to build that,” Bruce continued. “Now it feels more like I’ve become a project manager, directing my crack team of AI coders and letting them work out the details.”

Bruce speculates that his coding skills “may not be needed at all” once AI vibe-coding truly takes off, with AI agents “scouring the web for minor frustrations it can solve autonomously.”

Perhaps, but I think it’ll be awhile before an AI devises a solution as unique and ingenious as Bruce’s LEGO brick finder. Go check it out.