A coder used AI to rank which tech companies offer the best food at work. Check out the winners.



Lunches.fyi: Where Tech Rankings Are Served on a Plate

Move over, market cap—there's a new way to size up Silicon Valley.

Enter **Lunches.fyi**, a playful side project that ranks tech companies not by valuation or growth metrics, but by something far more universal: cafeteria quality. Built in roughly an hour by developer and self-described "coding prankster" Riley Walz, the site uses voice-dictated prompts to OpenAI's Codex to scrape publicly available tech company menus, then leverages AI to categorize and score meals on flavor, variety, and nutritional value. Walz estimates the same build would have taken 20+ hours using traditional coding methods.

The early results delivered a few surprises. NVIDIA, fresh off its AI-fueled stock surge, also topped the lunch leaderboard—serving everything from truffle mushroom pizza to nutrient-dense leafy greens that seem to mirror its upward trajectory.

But the project also underscores a familiar, critical lesson: **AI is only as good as the data it's built on**. When Replit CEO Amjad Masad publicly questioned his company's unexpectedly low protein ranking, Walz dug in—and quickly uncovered a bug. Missing nutrition fields had defaulted to zero, artificially deflating scores.

Once corrected, the rankings shifted.

It's a lighthearted experiment, sure. But tucked inside the fun is a sharp, timely reminder: without clean, complete data, even the most sophisticated AI can serve up misleading results.

Curious where your favorite tech company lands? Pull up a chair and browse the rankings at [Lunches.fyi](https://lunches.fyi). 🍽️

Post a Comment

Previous Post Next Post