I don't watch F1, so I built a quiz where the AI isn't allowed to grade me
This is a submission for the Sanity Challenge, Path One: Ship an Agent That Queries Real Content What I Built I'm not an F1 person, but I've wanted to get into it for a while. Pole, DNF, sprint, undercut: I knew the words existed and not much more. So for this challenge I built the thing I actually needed, a quiz that teaches you Formula 1 one race at a time. It's called Pit Wall. You pick a 2026…
This is a story about a creator who decided to build a quiz to learn about Formula 1 racing. The creator, who isn't interested in F1, wanted to gain knowledge about the sport. They created a game called Pit Wall, where the player picks a 2026 Grand Prix and a tyre option, and then answers five questions about the race. After every answer, the game displays the finishing order and explains the F1 terms used in the questions.
The unique feature of this quiz is that the AI writes the questions but is not allowed to grade the player's answers. The AI writes the questions and provides the GROQ query that answers them, while the server keeps the AI's opinion of the answer aside. The creator tested the AI's accuracy by comparing it to the same model without access to the data, finding that it got 16 out of 30 questions right while random guessing would have gotten 7 or 8 of the 30 correct.
The creator used 6,254 documents of real results from F1DB, including races, results, drivers, teams, circuits, and championship standings from 2014 to round 15 of 2026. They created two Sanity Context endpoints, one for facts and numbers and another for what the words mean. The creator imported nine Wikipedia articles to build the Knowledge Base and resolved a conflict related to the bonus points for fastest lap lap, which was scrapped from the 2025 season.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
