The hardware has been there since 2019.
When Apple started shipping AirPods with motion sensors in 2019, most people never noticed how capable those sensors were.
Every modern pair of AirPods contains a gyroscope and an accelerometer. Apple designed them for features like head tracking and spatial audio, but the hardware is capable of much more.
Kinapod uses those same sensors for sports tracking. Today it works with:
- AirPods Pro (2019 version and newer)
- AirPods (3rd generation and newer)
- Beats Fit Pro
No custom hardware.
No modifications.
Just software.
A different way of thinking about AirPods
Don't think of AirPods as earbuds. Think of them as wireless IMUs.
Most applications use AirPods exactly as Apple intended: worn in the ear. Kinapod starts with a different assumption. Instead of treating AirPods as earbuds, I treat them as tiny wireless motion sensors.
That means they don't have to stay in your ears. They can be attached to sports equipment, mounted on a shoe, embedded inside a flying disc, fixed to a racket, or placed anywhere motion needs to be measured.
The practical trick that unlocked this, found by trial and error, lives in the iPhone's AirPods settings: turn off Automatic Ear Detection, and fix the microphone to one specific bud instead of automatic. Without those two settings, iOS assumes the pod isn't in use and the motion stream doesn't behave.
To the best of my knowledge, this is the first platform built around using AirPods this way.
Once you stop thinking of AirPods as headphones, they become something else entirely: a compact wireless IMU that millions of people already own.
50 Hz is enough
Apple exposes real-time motion data from AirPods at 50 Hz. It wasn't always like this: for years the stream was capped at 25 Hz, and a recent iOS update doubled it.
At first glance, that doesn't sound particularly impressive. Dedicated sports sensors often sample at a few hundred hertz, and that's generally considered necessary for high-performance motion tracking.
As it turns out, that isn't always true.
Using the same 50 Hz signal, I've already built a running footpod, measured flying disc spin well beyond the gyroscope's saturation limit, estimated release speed, recognized movement patterns, and computed detailed sports metrics in real time.
The running footpod estimates treadmill pace with around 1% error.
What's even more exciting is that this result was achieved without training any machine learning models and without tuning the algorithm on recorded running data. The estimate comes purely from signal processing, physics, and carefully designed algorithms. I believe there's still significant room for improvement once machine learning is introduced.
The sensor usually isn't the limiting factor. Modern consumer devices already produce remarkably useful motion data.
The difficult part is figuring out what information is hidden inside that signal. That's where most of the work happens, and that's the part that stays behind the curtain.
Surprisingly long wireless range
One of the biggest surprises while developing Kinapod was wireless range.
In my testing, an iPhone maintained a stable connection to an AirPod at around 60 meters in line-of-sight conditions.
Even more interesting, paired AirPods communicate with each other over Apple's proprietary wireless link. A second pod extends the effective range and, more importantly, gives seamless reconnection to the phone if the link to one pod drops. In practice it behaves like a tiny mesh network.
That's far beyond what most people expect from a pair of earbuds.
Built for real-time feedback
Sports tracking only works if the data arrives fast enough.
In my testing:
- Typical end-to-end latency is below 150 ms
- Motion data continues streaming in real time without noticeable throttling
- I haven't observed software-imposed limits on accelerometer values during high-acceleration motion
- If the wireless connection is briefly interrupted, the stream automatically recovers once communication resumes
The goal isn't collecting data after your workout. It's giving feedback while you're still moving.
High-speed motion
One of my first questions was whether AirPods could keep up with fast sports.
They can, though not the way I first assumed.
The built-in gyroscope saturates at roughly 666 RPM. That ceiling is actually on the higher end for consumer gyroscopes, so it's expected behaviour for this class of sensor rather than a weakness of the AirPod. It is a hard hardware limit, though: past it, the gyro simply reports its maximum.
The accelerometer is what gets past that cap. Even at only 50 Hz, it still resolves rev rates well above the gyro's ceiling, and that is what makes rapidly spinning equipment such as flying discs measurable at all.
For many applications, the sensor performs considerably better than most people would expect from a pair of earbuds.
It actually started with running
Kinapod wasn't originally about frisbee.
Years before Kinapod Disc, I was experimenting with using AirPods as a running footpod.
The question was simple. Could a pair of earbuds replace dedicated running hardware?
Surprisingly, yes.
It almost didn't happen, though. For years the motion stream was capped at 25 Hz, which wasn't enough samples per stride for a footpod, and the project stalled. About a year ago, when I dusted it off, iOS 26 had raised the rate to 50 Hz, and that's when it got serious.
The result was the treadmill footpod described above: pace to about 1% plus full running dynamics, before any machine learning.
That project eventually led me to a much bigger question. If AirPods can understand running, what else can they understand?
Beyond frisbee
Disc sports were simply the first public demonstration of the platform. The app already covers disc sports, bowling, and running.
The same technology extends further: tennis, table tennis, pickleball, badminton, golf, hockey, lacrosse, skateboarding, track & field, strength training.
The only real limitation is where you can securely attach the sensor, and what meaningful information can be extracted from its motion.
In many cases, that's a software problem rather than a hardware problem. The hardware already exists. The challenge is discovering what it can measure.
Patent-pending
The core of Kinapod is covered by two U.S. provisional patent applications filed in 2026.
Free-placement inertial measurement. Every dedicated sports sensor sits at an engineered, factory-known position inside its equipment, and its algorithms lean on that known placement. Kinapod assumes the opposite: an earbud press-fitted snug through a pinched hole at the centre of a disc, dropped into a bowling ball’s thumb hole, or tucked under shoelaces sits at a position and orientation that are free, unknown, and unrepeatable. The filed methods make every measurement invariant to that free placement (the sensor calibrates itself from the equipment's own motion), and where free placement creates artifacts, they become signal: the off-centre wobble of a spinning disc is precisely what lets Kinapod read spin far beyond the gyroscope's hardware ceiling.
Acoustic Doppler speed from consumer audio. The same earbud-and-phone pair doubles as a Doppler speed instrument: one device emits a reference tone, the other listens, and the frequency shift of what it hears gives the object's speed. It is the physics-classroom "Doppler ball," rebuilt entirely from electronics people already own. Either device can take either role: the earbud can emit from inside a flying disc while the phone's microphone listens, or a loudspeaker can emit while the earbud's microphone rides the moving object. The enabling work was making a tone survive the wireless audio codec at all.
Both applications describe the general case, covering any sport, any moving object, and any consumer device with the right sensors, not only the features shipping in the app today.
My vision
I don't think athletes should have to buy a different sensor for every sport.
The sensors already exist.
Phones.
Earbuds.
Watches.
Wearables.
My goal is to figure out what those sensors can actually measure.