What Komori Can't Do (And Why We'll Say So)

The Limitations Page Nobody Writes

Every sleep gadget has a marketing page. Almost none of them have a page listing what the thing can't do.

That's not an accident. Limits are bad for conversion. So the category fills up with devices that promise to "detect" conditions they cannot detect, "monitor" things they merely estimate, and rate your night out of 100 using math nobody will show you.

We're building Komori in that category, so here's our limitations page. This is the honest list of what Komori is not designed to do, written before launch, in public, where you can hold it against us later.

None of this is an apology. A device that does a few things well and says so is more useful than one that claims everything and explains nothing. We'd rather you know the shape of the tool.

Key Facts

• A sleep study records brain waves, eye movement, chin and leg muscle activity, heart rhythm, airflow, chest and belly effort, and blood oxygen — all of it wired to you (StatPearls)
• In-lab polysomnography is the accepted reference standard for diagnosing obstructive sleep apnea (Journal of Clinical Sleep Medicine, 2017)
• Consumer sleep technology should not replace validated diagnostic instruments, per the American Academy of Sleep Medicine (JCSM, 2018)
• Contactless sensing has physical limits: accuracy depends on distance, obstacles in the signal path, body orientation, and motion (Sensors, 2023)
• One night is a bad sample. Across 1,041 people and 107,144 nights, 3 to 5 nights were needed for a reliable weekly sleep estimate, and 11 to 18 for monthly night-to-night variability (SLEEP Advances, 2022)

1. It Can't Diagnose Anything

Komori is being built as a general wellness device. It is not a medical device, and it will not detect, diagnose, screen for, or treat any medical condition. Not sleep apnea, not insomnia, not a neurological condition, not anything else.

This isn't modesty. It's a description of what the tool is.

The American Academy of Sleep Medicine put the general principle plainly for consumer sleep technology as a category: data from these devices "should be considered in the context of a comprehensive sleep evaluation and should not replace validated diagnostic instruments" [^1]. That's the right standard, and we hold ourselves to it.

If something in your nights worries you, the answer is a clinician. A device that shows you patterns is a conversation starter with your doctor, never a substitute for one.

2. It's Not an Emergency Alerting System

Komori is not a medical alert device, and it should never be used as one. Do not deploy it as the safety net for someone whose life depends on a fast response.

That constraint matters most for the people who most want to hear otherwise — families caring for someone at night, who are exhausted and would love a machine that watches while they sleep. We understand that pull. We're not going to exploit it.

A wellness device that logs patterns overnight is a fundamentally different product from a clinical alarm system, which needs the reliability engineering and redundancy that a life-safety role demands. Anyone who blurs that line is selling you a feeling, not a system. If you need medical alerting, get a system built for medical alerting.

3. It Can't Replace a Sleep Study

This is the limit people find hardest to accept, because home devices have gotten impressive at the things they can do. So let's be specific about the gap.

An in-lab sleep study — polysomnography — records many streams of physiology at once, most of them requiring physical contact with your body: EEG electrodes on your scalp for brain waves, EOG sensors beside your eyes, EMG on your chin and legs for muscle activity, an ECG lead for heart rhythm, a nasal pressure transducer and a thermal flow sensor for airflow, elastic belts around your chest and abdomen for breathing effort, and a pulse oximeter on your finger for blood oxygen [^2].

Nearly all of that is physically unavailable to a device across the room. Brain waves are microvolt-scale electrical signals measured at the scalp. Airflow is measured at your nose and mouth. Blood oxygen is measured by shining light through your tissue. No contactless sensor sees any of those. That's not an engineering problem waiting for a better algorithm. It's what the signals are.

Which is why in-lab polysomnography remains the accepted reference standard for diagnosing obstructive sleep apnea, and why the AASM's diagnostic guideline explicitly recommends against using clinical tools, questionnaires, and prediction algorithms to diagnose it in the absence of a proper study [^3].

If you've had a sleep study and want help reading it, we wrote a plain-English guide to sleep report metrics. That post exists precisely because the study is the thing that answers the medical questions, and Komori isn't.

4. It Can't Tell You Why

Komori is being designed to describe what happened, not to explain it.

Once it ships, say a night shows a run of restlessness around 3am and a bed exit shortly after. That is what the pattern was. Why is not in the data. It could be a full bladder, a bad dream, a partner rolling over, a noise outside, alcohol at dinner, pain, stress about a meeting, or nothing identifiable at all. The sensor sees the movement. It does not see the cause.

Plenty of products will happily narrate a cause for you anyway — "you were stressed," "your recovery is low." That narration is generated, not observed. We'd rather show you the pattern, timestamp it, and let you supply the context, because you're the one who knows whether you had a glass of wine at nine.

The useful version of this is boring and works: notice a pattern, form a hypothesis, change one thing, see whether the pattern moves. That's it. That's the method.

5. One Night Tells You Almost Nothing

If you buy a sleep device, look at one night, and draw a conclusion, you've made a measurement error.

Sleep is noisy. Any single night reflects that day's caffeine, stress, room temperature, schedule, and randomness. The signal you actually want lives in the average and in how much you vary around it — and both take time to see.

There's good data on how much time. A study of 1,041 adults across 107,144 nights found that estimating someone's habitual weekly sleep duration took 3 nights for a "good" estimate and 5 for a "very good" one. Monthly estimates took 5 and 10 nights. And measuring variability — how much your nights differ from each other — took 11 to 18 nights for a monthly picture ¹.

That's the honest framing for any sleep tracker, ours included. The first night is nearly meaningless. The fiftieth is where it gets interesting. Anything that hands you a dramatic verdict on night one is performing, not measuring. We've written about how that performance backfires in our post on orthosomnia — the anxiety that sleep tracking itself can create.

6. There Won't Be a Sleep Score

Komori will not give you a number out of 100 that claims to summarize your night.

We've made the full argument in why we're not shipping a sleep score, so the short version here: a single score takes several unrelated measurements, weights them with judgment calls no vendor publishes, and outputs a figure precise enough to feel authoritative and vague enough to be unfalsifiable. You can't act on an 82. You can act on "you spent most of the night on your back, and you didn't on the nights you felt better."

Refusing to ship a score costs us. Scores are the single most screenshot-friendly feature in this category. We're not shipping one anyway.

7. The Physics Has Limits, Too

Even inside its actual scope, contactless sensing isn't magic. A 2023 systematic review of contactless sleep-sensing technologies across 54 studies and 15 sensor types catalogued the recurring constraints: accuracy depends on the distance and position of the sensor relative to the person, on obstacles in the signal path, and on motion and body orientation ².

Translated into your bedroom: placement will matter, and a device that measures a room is measuring a room, including everything else in it. We'd rather say that now than have you discover it later and feel misled.

Where Komori Fits

Here's what's left after all of that, and we think it's a good list.

Komori is being built as a contactless device that senses from above the bed — no camera, no wearable, nothing touching you — designed to track a specific set of things night after night: your sleep position, how much you move and how restless a night was, when you got out of bed, what your room was doing (temperature, humidity, CO2, light, and noise), and sound events.

The value in that isn't any single reading. It's the baseline. When it ships, someone might notice that their restless nights cluster with the nights their bedroom ran warm, or that their time on their back has crept upward over a season, or that a change they made three weeks ago actually moved something. None of that is a diagnosis. All of it is real information about your own life that you currently don't have.

The privacy design follows the same logic — no camera in the bedroom, and clear limits on what leaves your house. That's covered on our privacy-first page. And the research we track that informs the work is on the research page.

Why We're Publishing This

Because the alternative is worse.

The sleep tech industry has spent a decade training people to distrust it — overpromising, then quietly walking things back in a footnote or a firmware update. We'd rather set expectations low enough to be true, then be judged on whether we met them.

There's also a practical reason. If you know exactly what a tool can't do, you use it better. You stop expecting a diagnosis from a movement sensor. You stop reading one night as a verdict. You start asking the questions the data can actually answer, and you bring the rest to your doctor, which is where it belonged the whole time.

That's the deal we're offering: a narrow, honest instrument that measures a handful of things carefully, and a company that will keep saying out loud what it doesn't measure. If we ever quietly delete this page, hold it against us.

---

Footnotes

1. Lau TY, Ong JL, Ng BKL, et al. "Minimum number of nights for reliable estimation of habitual sleep using a consumer sleep tracker." SLEEP Advances, 2022. ↩

2. Boiko A, Martínez Madrid N, Seepold R. "Contactless Technologies, Sensors, and Systems for Cardiac and Respiratory Measurement during Sleep: A Systematic Review." Sensors, 2023. ↩