You've probably seen AI tools that generate text, images, or code. But analyzing a physical movement like a tennis stroke — where body position, timing, and racket path all matter at once — is a very different kind of problem. Here's how Stroke Coach does it.
Step 1: Video stays on your device
When you upload a video to Stroke Coach, the first thing that happens is probably not what you'd expect: the video never leaves your phone or computer. Instead, your browser extracts a set of evenly-spaced frames directly on your device — think of it as pulling out 10 snapshots that cover the full motion from setup to follow-through.
These frames are what get sent for analysis. The original video file stays on your device and is never uploaded to any server. This matters both for privacy and for speed — sending a few lightweight images is much faster than uploading an entire video file.
Step 2: AI reads the frames like a coach watches video
The extracted frames go to an AI vision model that has been given deep context about tennis technique. It's not a generic image classifier — the system prompt is specific to the stroke type you selected (forehand, backhand, or serve) and tells the AI exactly what to look for.
For a forehand, it's evaluating grip type, unit turn, racket takeback, contact point position, follow-through direction, footwork, and balance. For a serve, it's looking at ball toss placement, trophy position, racket drop, contact height, pronation, and landing mechanics. Each stroke has its own coaching lens.
The AI scans all the frames and identifies which specific frame best shows each issue. This is what makes the feedback concrete rather than generic — it's not just saying "improve your follow-through," it's pointing at the exact moment where your racket stops short and explaining what should be different.
Step 3: Structured feedback, not a wall of text
The AI returns its analysis in a strict format: an overall score out of 10, a list of specific strengths it observed, and the top three things to fix — each with a title, a detailed description of the issue, and a concrete drill to address it. Every fix includes a reference to the frame where the issue is most visible.
This structure matters. A human coach giving verbal feedback might cover 10 things in two minutes, but you'll remember maybe two of them. The structured format means you get a report you can take to the court and work through one fix at a time.
How is this different from a human coach?
It's not a replacement — it's a different tool. A human coach can rally with you, feed you balls, adjust your grip with their hands, and read your body language to know when you're frustrated. AI can't do any of that.
What AI does well is the tedious, frame-by-frame analysis that even good coaches don't always do. Most lessons are an hour long, and the coach is watching in real time — they're not pausing video and measuring your racket angle at contact. AI does exactly that, consistently, every time, and returns a written report you can reference weeks later.
The sweet spot is using both. Film yourself during practice, get an AI analysis to see what's actually happening in your strokes, and bring those insights to your next lesson. Your coach will appreciate that you already know what to work on.
What about privacy?
Your video is processed entirely on your device. Only the extracted frames are sent for analysis, and they're not stored afterward. There's no video library, no cloud storage, no account data being mined. The frames go in, the analysis comes back, and that's it.
Curious what AI sees in your stroke? Try it — your first 3 analyses are free.
Analyze my stroke — free →