While other sports camera companies were building robots, we were doing something different. We were filming real games with iPhones and teaching high school students how to broadcast them.
It turns out that was the smartest thing we could have done.
The bet the industry made
The dominant approach in consumer sports cameras is hardware first. Build the device, then figure out how to make it smart.
The result is a generation of cameras that film games like security footage. Wide, flat, and far away. They see the whole field and miss the moment.
And there's a harder problem underneath: an AI camera is only as good as the footage it learns from. A camera can't learn good filming from its own mistakes.
The bet we made
Apollo Streams spent years filming games before we ever built hardware. Football, volleyball, baseball, basketball. Real games, real weather, real Friday night lights.
Since then, families and schools have used our system to broadcast thousands of games. We watched much of that footage back, and every review exposed another real-world problem to solve. Brutal light, sudden weather, unpredictable kids, scoreboards of every shape: every product decision we've made traces back to something we saw go wrong at a real game.
The part that teaches the AI is smaller and more deliberate. We ran a junior broadcasting program that taught high school and college students how to film sports the right way. When to stay wide. When to zoom. How to read a play before it develops. Every student made mistakes, got coached through them, and got better.
That program left us with something we've never seen another camera company build: a human-reviewed library of game film. Every game in it was critiqued with the students who filmed it, so the library holds the same plays filmed well and filmed poorly, along with the corrections that turned one into the other.
Today that library holds:
- 100+ complete American football games across three full seasons
- 90+ volleyball games
- 20+ baseball games, growing every season
- More basketball games than we can count
The broad archive shows us how real games unfold. The coached library shows the AI how those games should be filmed.
An AI model needs to learn two things: what to do and what to avoid. We have both. We spent years coaching humans through the learning curve. Now we're coaching the machine.
Why football is the real test
Football is the hardest sport in consumer sports filming, and it keeps exposing the limits of tracking cameras.
The ball is small. The camera sits far from the action. And every single play forces a decision within about two seconds of the snap: is this a run or a pass, and where should the camera go?
Most consumer systems sidestep that decision by capturing the whole field at once. That preserves coverage, but it trades away closeness and detail. When the camera has to see everything, your kid is small in the frame. Getting close requires zoom, and zoom forces exactly the split-second, read-the-play decisions that are hardest to automate.
BallHawk can zoom, and the coached library is what teaches it where to point. That changes what's possible.
More than five sports
BallHawk launches with AI game tracking for football, basketball, volleyball, baseball, and softball, each backed by years of real footage from our library.
The system reaches well past those five. Sports that happen in a fixed space, like wrestling, pickleball, and tennis, film great with a set camera. Athlete tracking is designed to follow one player across many sports. And live streaming, instant highlights, and on-screen scoreboards for more than 20 sports come built in either way.
We're not guessing how AI should film your kid's game. We taught it the same way we taught our students. One game at a time.
Why this matters to you
If you're a parent, it means the camera finally follows the game the way you would film it yourself, so you can watch your kid instead of a screen.
If you're a coach or league, it means broadcast-quality footage without a broadcast budget.
And if you're wondering whether BallHawk will really film sports better than what's out there, this is why we believe it will. The hardware matters, but the library is what teaches the AI how sports are actually supposed to be filmed. Watch actual game footage here.
BallHawk's intelligence lives in the app, and updates are included with your subscription, so it keeps getting better after you buy it. Because the camera is the iPhone you already own, those improvements arrive as software instead of a new multi-thousand-dollar device. The system you have is the system that improves.
That real-world development process is a story of its own: Why You Can't Build a Great Sports Camera From an Office.
The part you came for
You came to the field to watch your kid play, not to study camera work.
We already did that part.
Never miss the moment again.
BallHawk launches on Kickstarter this fall. Join the early access list to get launch pricing before anyone else.
