Series A · circadian performance systems V3LA One — clinical validation underway
Platform

One sensor, one model, one instruction.

V3LA is a closed loop: a device that measures circadian phase continuously, a model that infers where your internal clock actually sits, and an app that turns that estimate into three or four decisions a day. Everything below is one of those three parts.

01 — Hardware

V3LA One

A titanium ring with no screen, no haptics and no optical heart-rate sensor. Dropping PPG is what buys the battery life and the thermal stability — and PPG was never the signal we needed.

Two thermistors sit against the finger with a known thermal path to ambient, so we can separate a genuine drop in distal temperature from a cold corridor. A 6-axis IMU supplies rest-activity and, at night, an HRV proxy from ballistic motion.

Sensors2× thermistor, 6-axis IMU
Sampling1 Hz continuous
Battery9 days typical
RadioBLE 5.4
Mass / rating3.1 g · 5 ATM
Sizes6 – 13 US
What it deliberately doesn't have
No optical heart-rate sensor
PPG is the single largest power draw in a ring and contributes almost nothing to phase. Removing it bought the nine-day battery.
No screen, no haptics
Nothing on your hand should buzz at 03:00. The phone is the interface; the ring is only a sensor.
No daily charging
A tracker you take off to charge is a tracker with holes in exactly the window that matters.
No cloud requirement
The device pairs to one phone and works fully offline. Backup is opt-in.
02 — Model

The estimator is the product.

Anyone can plot a temperature curve. The hard part is inferring phase from a signal that is contaminated by ambient temperature, activity, alcohol, illness and the fact that people take rings off. Three things make it work.

Personalised oscillator

A limit-cycle model with per-user amplitude and intrinsic period, fit by particle filter. Your τ is not 24.0 hours and the model doesn't pretend otherwise.

Light history

Phase response to light is nonlinear and direction-dependent. We integrate ambient exposure from the phone and treat it as the primary control input, not context.

Honest uncertainty

Every estimate ships with a confidence interval, and the app says less when the interval is wide. Silence is a feature in a product people trust with a night shift.

On-device inference

The whole loop runs on the phone in under a second. Nothing needs to leave the device for the product to work, which also settles most of the privacy conversation.

Example plan · night 2 of 4
What the app actually shows: six decisions, in order, with the reason attached.
08:20 Sleep now — anchor block, 6 h 40 m. Your temperature minimum lands at 11:40.
15:00 Stay dark 40 min — light now would drag your phase the wrong way.
20:30 Bright light, 30 min — the one intervention that moves you tonight.
21:40 Caffeine window opens — cut-off 02:00 to protect tomorrow's block.
03:10 Predicted low — reaction time bottoms out for roughly 90 minutes.
07:00 Sunglasses home — the commute is the biggest uncontrolled light dose of the day.
03 — Software

Plans, not dashboards

You give V3LA your roster once — typed, imported from a calendar, or pulled from a workforce system. From then on the app works backwards from each shift to the decisions that change how it feels.

Anchor sleep window
One protected block, placed where your phase will actually let you sleep — not where the clock says night.
Light prescription
When to seek bright light and when to avoid it, in twenty-minute windows, with the direction of the shift it produces.
Caffeine and nap timing
Placed to protect the anchor block rather than to maximise alertness right now.
Impairment forecast
The hours in your next shift where reaction time is predicted to bottom out — useful to know before you're in them.

Integrations

Apple Health

Reads sleep and workouts, writes phase-adjusted sleep windows back.

Calendar (ICS)

Any roster that publishes a feed becomes a schedule V3LA can plan against.

Workforce systems

Kronos and Workday connectors for enterprise deployments, read-only.

Research export

Raw 1 Hz series and phase estimates as CSV for collaborating labs.

See the evidence behind it.

Validation methods, pilot results and the phase-estimation error we publish against.

Read the science Join a pilot →