Analyze your Gait

Give us one clip, and we'll give you a plan you can trust.

Past Analyses

No analyses saved yet.

Already filmed one? instead.

Record with your camera

Prop your device ~3 m away, full body in frame. Same analysis as an upload.

Don't save my clip. We normally keep it for up to 30 days to check the analysis is accurate, then delete it. Tick this and it is deleted the moment your results are ready.

Original Video

Movement Analysis

Not medical advice. Sauntle is a training and prevention tool, and we are not medical professionals. These results come from pose estimation on a phone video and have real measurement error. They are for education, not a substitute for professional care. If you have pain or an injury, talk to a qualified professional.

What the analysis looks like

Every clip comes back the same way. Each metric is graded against movement-specific norms, and any flagged pattern is explained instead of left as a bare number.

A score out of 10, with every metric behind it shown
One plain-language finding per metric, flagged or not
Exercises matched to what was flagged, and shoe options on running clips

Sample analysis Running

Cadence 176 spm

Knee angle 168°

Pelvic drop 6.8%

Trunk lean 7°

8.4 / 10

One flagged pattern, with a fix

Objective movement analysis, from one phone clip

Sauntle reads the movement patterns that shape running form — cadence, knee flexion, pelvic drop, trunk posture, left–right symmetry and pronation — from a single video of you walking, running, or landing a jump. No wearables, no markers, no appointment.

Why it exists

Gait labs produce numbers. Everywhere else you get opinions: a form cue from a video, a shoe picked by feel, a guess at why one hip aches after long runs. Sauntle puts a measurement where the guess was, using footage you can film on a phone propped against a water bottle.

How a number is judged

Each metric is graded against reference ranges for the movement you filmed — walking, running and jump landings each have their own. Per-frame readings are combined with robust statistics rather than a plain average, so a handful of bad tracking frames cannot drag your result with them.

What it will not do

It does not diagnose, and it does not fill in gaps. Anything the camera could not see is reported as not measured instead of invented: a side view cannot show pronation, and a clip cut off at the waist returns leg metrics only.

What gets measured

Cadence Steps per minute, counted from detected foot strikes rather than assumed from the length of the clip. Skipped, with a reason, when the clip is too short or the feet are not clearly visible.
Knee flexion Peak knee angle on each side, computed in 3D so a leg angled toward the lens is not read as straighter than it really is.
Pelvic drop How far the hips tilt through the stride, as a percentage of your own torso length so the reading does not depend on how far you stood from the camera. Needs a front or rear view.
Trunk lean Forward lean of the torso against vertical, from a side view — the posture metric most often behind a heavy, braking stride.
Stride symmetry How closely your left and right sides mirror each other, as a percentage. Persistent asymmetry is what turns a one-off niggle into a pattern.
Pronation Ankle roll through stance on each foot, from a front or rear clip. It is also what the shoe suggestions are built on.

How it works

Upload or record

Drop in a clip filmed from the side or from behind, or record one in the browser with the on-screen guide. A few seconds of steady, full-body footage is enough, and the clip is checked for length, lighting and framing before anything is analysed.

We track your joints

Computer-vision pose tracking locates your hips, knees and ankles frame by frame across the whole clip. Frames where the pose is anatomically implausible are thrown out rather than averaged in, so a glitched frame never becomes a finding.

Get results you can use

Every metric is graded against movement-specific norms and then explained: what the number means, whether it is worth acting on, and which exercises target it. Sign in and each analysis is saved, so you can watch a number move over weeks instead of guessing.

Metrics tracked per clip

Movement types supported

<60s

Typical time to results

Built by the runner it was too late for

Sauntle is made by two juniors at Gunn High School: one who lost a year of running to an injury nobody saw coming, and one who looked at that year and saw a measurement problem waiting to be solved.

Tarun Nikhil

Co-founder · Junior, Gunn High School

Tarun finished 11th in the Mile at Nike Indoor Nationals — one of the fastest high-school fields in the country — and then spent the next year injured instead of racing. A season became a year of rehab, and the hardest part was that nothing in his training told him it was coming: splits, mileage and shoes all looked fine right up until they didn't.

The things that would have shown the pattern building — how his hips held up late in a workout, whether his stride stayed even as he tired — are measured routinely in gait labs and almost never available to the runners who need them most. Sauntle is the tool he wanted before that year, not after it: the same measurements, from a clip filmed on a phone at the side of a track.

Aarush Rachakonda

Co-founder · Junior, Gunn High School

Aarush works on the half of Sauntle that is a maths problem wearing a tracksuit. He competes in mathematics and programming contests and spends his own time on machine learning, which is almost exactly the stack this product runs on: pose estimation frame by frame, robust statistics so one glitched frame never becomes a finding, and scoring rules that still have to behave at the edges.

He is the reason a reading is graded against a defined range rather than a hunch, and the one arguing for the boring, defensible version of a metric when the flashy one would have been quicker to ship.

Get in touch

Questions, a metric that looks off, an exercise you would like added — we read everything.

Email us

One address, read by the people who build this.

gaitai.athle@gmail.com

Reporting a wrong reading

Tell us the movement type you picked, which metric looks wrong, and roughly what the clip shows — camera angle, distance, indoors or out. That is usually enough for us to reproduce it.

Your clips

Ask and we will delete any clip saved to your account. Signed out, a clip is deleted the moment your results are ready unless you tick the box to share it.