Hi Product Hunt — I'm Sam, and I built Court Rally for my nine-year-old, who trains on a table tennis robot an hour a day. The problem was never motivation. It was that between weekly lessons we had no plan and no feedback. I'd film him, watch the clip, and not know what I was looking at. His coach could tell in two seconds — but only on lesson day. So I built the thing I wanted: a plan for the week that a coach would recognize, and a read of the film that says the one thing to fix, in words I can repeat at the table. Then his coach asked if he could see the reads too, and the app grew a coach side: lessons, proposals, a schedule, a club desk. Two design choices I'd defend: - The read says what it can't see. A checkpoint the frames don't show is marked "can't tell", never guessed. We'd rather be useful than confident. - The human coach stays. This isn't a replacement for a coach; it's what happens between lessons, and it hands the coach better information. It covers table tennis, tennis, badminton and pickleball. The four sample reads at courtrally.fun/sample are real clips, real output, with the frames shown. I'd genuinely like to hear where the reads are wrong — that's how the prompt got as good as it is.

Court Rally

An AI coach that keeps your real coach in the loop

An AI coach that keeps your real coach in the loop