Sleep Grove is a personalised 90-minute wind-down — light tapering, circadian-aware audio, optional breath pacing — built around a single insight from our cohort: clock discipline matters more than theatre. Most nights it asks very little of you, and quietly turns down the room.
A representative sequence on a target sleep onset of 23:00. Timings shift with your circadian profile, sunset, and the last seven nights of actigraphy.
If smart bulbs are connected, the model tapers brightness and shifts colour temperature toward 2300 K. If not, you get a single nudge: "would now be a good moment for the lamps?"
The model proposes a screen-cool-down — not enforced, not gamified, simply offered. Cohort data shows that suggestion-with-rationale outperforms enforcement at 14 weeks.
Delivered between 21:15 and 21:50 in the cohort the effect was real; outside that window, negligible. The model only proposes this segment if the window is open for you tonight.
Recorded in Cumbria, the Mendips, and a closed library in Kings Cross. We chose not to generate audio: the small loss of personalisation is worth the gain in not training on bedtime data.
The app simply sleeps. The next signal it sends you is a written one-paragraph summary on Sunday — and only if there is something worth saying.
The Bamboo Grove sleep cohort followed 1,400 UK and Ireland participants, aged 22 to 71, instrumented with consumer-grade actigraphy and weekly self-report, over fourteen months. Half received Sleep Grove's personalised wind-down; half received a strong, fixed baseline. The principal findings:
| Outcome | Baseline | Sleep Grove | Δ |
|---|---|---|---|
| Sleep onset latency | 22 min | 14 min | −8 min |
| Subjective rest (1–7) | 4.1 | 4.9 | +0.7 SD |
| Wake-after-sleep-onset | 34 min | 28 min | −6 min |
| 14-week retention | 41% | 72% | +31 pp |
| Caffeine-cutoff adherence | 17% | 64% | +47 pp |
Full methodology, dropout analysis, and pre-registration in the cohort report — request via the contact page or read the long-form journal entry.
The most powerful variable in the entire study was not any audio scene, breath pattern or light protocol. It was the cutoff time for caffeine. Coaching users to a hard 14:00 cutoff produced more improvement, on more outcomes, than any other intervention we tested. Sleep Grove encodes that finding into its very first calibration question — and from there into a quietly persistent, written nudge.
Sleep tech, we now believe, needs less theatre and more clock discipline. That is what this ritual aims at.