A new side project, Lunches.fyi, takes the AI tooling craze in a lighter direction, ranking tech companies not by market cap, but by cafeteria quality.
It was built by coding prankster Riley Walz in about an hour using voice-dictated commands into OpenAI’s Codex (a task he says might previously have taken 20+ hours).
The site scrapes publicly available tech company menus and uses AI to categorize and score meals. Early results crowned Nvidia a surprise leader, serving everything from “truffle mushroom pizza” to leafy greens worthy of its stock performance.
But the experiment also highlights a familiar truth: AI is only as good as the data it’s built on. Replit CEO Amjad Masad publicly questioned his company’s low protein ranking, prompting Walz to uncover a bug: missing nutrition data that had defaulted to zero.
Sorry for the protein slander! Some menus didn’t have nutritional facts, so the script assumed there were 0 grams of protein in all of Replit’s dishes. Just fixed. The joys of vibe coding…
— Riley Walz (@rtwlz) March 10, 2026
Fixing the error quickly changed the results. It’s a small, fun project, but a sharp reminder: AI breaks without quality data.
You can check out how tech companies rank, over that the website: Lunches.fyi.
Sign up for BI’s Tech Memo newsletter here. Reach out to me via email at abarr@businessinsider.com.
