Oh God, Vibe Coding on Smart Glasses Is a Thing Now

If there are two tech things you’re going to read a lot about in the coming year, it’s vibe coding and smart glasses, and lucky for you, you now get to read about both simultaneously. Thanks to a software engineer, Jake Ledner, the buzziest AI agent, OpenClaw, has now found a new friend in Meta’s flagship smart glasses, the Meta Ray-Ban Display. Instead of buying things on Amazon for you, though, OpenClaw is here to help you vibe code while you take a little stroll through Wall Street.

I’m building apps directly from my glasses 🤯
OpenClaw + Meta Ray-Ban Display pic.twitter.com/0WtBJ85gAK

— Jake Ledner (@jake_ledner) February 19, 2026

In a demonstration on X, Ledner shows how he connected OpenClaw, which is running on a Mac Studio in his apartment, to the Meta Ray-Ban Display, using OpenAI’s Codex tool to build apps with voice inputs. Because of the in-lens screen on the Meta Ray-Ban Display, Ledner can actually see the progress of the vibe coding session. For demo purposes, Ledner uses the setup to vibe code parts of a health-tracking app called “TrackGPT,†which sounds like exactly what you’d do with smart glasses that churn out apps for you.

Ledner even brings the whole thing a step further. Using his voice, he asks OpenClaw to code the feature but also to actually push the whole thing to a live app—a task it appears to complete in the demo. Welcome to the future of coding, I guess? As long as you’re okay with cranking out app slop.

While Ledner’s demo is pretty impressive from a technical standpoint, it’s not the only example of experimentation in cramming OpenClaw and Meta’s smart glasses together. Just last week, I covered how one software engineer hacked the Ray-Ban Meta AI glasses to buy things on Amazon by just looking at an object and asking the AI agent to add it to his cart. Again, the whole thing is interesting, but it probably isn’t something you should try for yourself right now.

However useful OpenClaw may be, it’s also a pretty huge security risk. For a more technical breakdown of the risks OpenClaw poses, you can read this Medium post from author Vishal Rajput, but the long and short of it is that OpenClaw—in order to do useful stuff—requires access to some of the most sensitive data, including passwords, browser history, and cookies, as well as files and folders on your machine. The security risks are so great that one cybersecurity firm, Palo Alto Networks, went as far as to say that OpenClaw constitutes a “lethal trifecta†of security risks.

In any case, people seem to be down to experiment with AI agents on smart glasses, regardless, so I would buckle up for even more mashing of OpenClaw and smart glasses in the future—for better and most likely worse.