BikeFit AI

Triathlon Fit Analysis Across 8–20 Pedal Cycles β€” Not a Single Frame

Generic bike-fit tools estimate. We compute measurement consistency on every analysis and show you a confidence score.

Knee angle (BDC) 143.1Β° Ideal
Measurement consistency 85% [8 cycles scanned]

Source: Holmes et al. 1994, Bini et al. 2011

Start analysis

The problem with generic fit tools

Most consumer bike-fit tools treat triathlon as one of several preset rider types. That often means aero geometry β€” high hands, forward saddle / steep STA, flat torso β€” gets scored with road-bike heuristics. The result can look β€œcorrect” on a single still while missing cycle-to-cycle consistency and triathlon-specific ranges.

How it works

1

Record on a trainer

Capture a side-view video while pedaling indoors. Stable framing improves BDC/TDC detection.

2

AI reads BDC/TDC across cycles

We locate bottom and top dead centre frames and aggregate angles over multiple pedal revolutions β€” typically in the 8–20 cycle range when the clip is long enough.

3

Report with confidence + references

You get joint feedback tied to published knee ranges and coaching consensus for aero posture, plus a consistency score so you can see how stable the measurement was.

Why this is different

Confidence score

Shows how many cycles were scanned and how much BDC knee angles varied (standard deviation). A high score means the reported angle is repeatable β€” not a lucky single frame.

Scientific references

Knee (BDC): Holmes / Bini / Ferrer-Roca. Hip, torso, elbow (aero): F.I.S.T. protocol and triathlon coaching standards β€” kept as separate evidence categories. Methodology β†’

Segment ratio

Shin/thigh pixel ratio at BDC is stored per session. Over time it supports personalized saddle/cleat context instead of one-size anthropometry assumptions.

Camera alignment warning

Off-axis camera angle is a common source of fit error. Calibration checks flag wheel/hub geometry that suggests the bike is not square to the lens.

Founder

Built by Onur Kapucu β€” triathlon coaching background combined with software engineering. BikeFit AI is a practical tool for measuring position on a trainer, not a marketing claim about replacing a full studio fit.

Pricing

Credit-based access for web analysis. Credits do not expire.

Analysis pack
€10

5 analysis credits Β· no time limit

Start analysis
AI + Coach Review
€30

1 Coach Review + 5 Analysis Credits

In addition to AI analysis, triathlon coach Onur Kapucu personally reviews your fit results and sends a custom video/written evaluation.

Request Coach Review

FAQ

Do I need a smart trainer?

A stationary trainer (smart or basic) is recommended so the bike stays fixed in frame. Bluetooth power/cadence is optional telemetry, not required for angle analysis.

How long does an analysis take?

A typical clip is ~20–40 seconds of pedaling after you’re on the bike. Processing on the analysis backend usually finishes within about a minute depending on length and load.

Which bike types are supported?

Road and triathlon / TT modes. Triathlon mode uses aero-specific torso, hip, and elbow guidance; road mode uses road ranges.

How should I film?

Side view, full body visible, camera roughly level with the bottom bracket, bike square to the lens. Mark hubs / BB / saddle in calibration when you want cm saddle and KOPS.

What is the confidence score?

It summarizes measurement consistency across detected pedal cycles (count + variability). Use it to decide whether to trust a single session or re-record with a longer, cleaner clip.