Baku
Apple Health · Oura · iPhone

An instrument,
not a scoreboard.

Baku models your sleep, recovery and biological age against your own history — not a population average. Every number it shows you is measured against the person you actually are.

Private beta, invited in small waves. No spam, and you can leave at any time.

BED 23:12 WAKE 07:29 DIP HRV SLP RHR LOAD SpO₂ 81 READINESS
Sleep
Training
Elevated stress

The Today screen · the day so far, lights out at 23:12

Precision

Every number is earned.

A score on your first day can only be a population average. It has almost nothing of you in it yet, however confident it looks.

Baku spends your first week learning your range — your HRV, your sleep architecture, your recovery rhythm. From seven nights on, every score is measured against you, and it keeps sharpening as your record grows.

That is what makes a number worth acting on: it is measured against you, not against everyone.

How it works

Your data, properly interpreted

Three steps, and the middle one is where the modelling happens.

1

Connect Apple Health or your Oura ring

Baku reads what your devices already record — and pulls your history, so the model starts with everything you have rather than from zero.

2

It learns your normal

Your HRV range, your sleep architecture, your recovery rhythm. Readiness appears after seven nights; the fuller model keeps calibrating for about three weeks, and tells you where it is the whole time.

3

One screen each morning

What last night did to today — readiness, sleep quality, training load and long-term trend, on a single dial. Wearing two devices is handled: Baku knows they disagree and corrects for it.

Biological ageA composite of VO₂max, HRV, resting heart rate and sleep — and which of them is holding you back.
Sleep architectureDeep, REM and light, with timing and regularity — not just hours in bed.
HRV & resting HRAutonomic balance, read against your own baseline rather than a chart norm.
VO₂maxCardiorespiratory fitness, estimated from what you actually did.
Training loadAcute against chronic, so a hard week reads as a hard week.
n=1 experimentsTest a change on yourself, and find out whether it actually moved anything.
What a year looks like

One person, twelve months in

A 34-year-old who trains four times a week. Every score here is produced by the same functions that run in the app — the composite is what its five components actually add up to.

Readiness · today
81

Recovered and ready to train. The load term is the one carrying it — the last block was absorbed cleanly.

The fingerprint · five components
HRVSLP RHRLOAD SpO₂
Last night
DEEP 18%
REM 21%
LIGHT 61%
Asleep7h 32m
Efficiency91%
Bed → wake23:12 → 07:29
Regularity (SD)±38 min
Biological age
29.4years

Against a chronological 34. VO₂max is doing most of the work; resting heart rate is the weakest of the four drivers.

29.4 YOU, 34 26.7 32.1

The band is your plausible range — it tightens as your record grows

Readiness · last 30 days
57 · HARD BLOCK, DAY 20
MEAN 78
81 · TODAY
Why Baku

Built by a researcher, for people who want the real answer

Showing numbers is the easy part. The harder problem is saying what they mean for you, on this day, given your history.

Calibrated to you

Every metric is read against your own distribution. A threshold fitted to someone else is a guess about you.

Measured, not assumed

Every value comes from something your devices actually recorded. Nothing is invented to make a screen look complete.

Private by construction

Your health data is never sold or shared, and the daily summary is written by a model running on your iPhone — not in a datacentre.

In Japanese folklore, Baku is the dream eater — a guardian that watches over sleep and turns what troubles you into strength.