Rhythm Fit
Your cycle dates and phase stay on your phone. We never send them to our servers, and we never sell your data.
A workout log that responds to your body. Free to start.
Get Rhythm Fit on the App Store
iPhone, with an Apple Watch companion.
Tap a question to open the answer here.
Wearables
Apple Watch, Apple Health, the Oura Ring and WHOOP today.
Four connections, and each one has a page of its own.
Everything a device sends changes how confident a suggestion is. None of it sets the weight on the bar — your logged sets do that. And Rhythm never shows a provider's own scores back to you; what you read is Rhythm's plain-language take on your day.
Overnight wrist temperature comes only from the newer Apple Watch models that carry the sensor; an older watch sends heart rate and sleep but no temperature. And a wearable is optional: Rhythm reads your training log first, so it works with nothing connected at all.
Yes: the Watch app runs a whole workout from your wrist, and Apple Health is where Rhythm reads your heart rate, sleep and recovery signals.
On the Watch you can log a set, change the weight or the reps with the crown, start, extend or skip your rest, and add an exercise from a catalogue that lives on the watch itself. It keeps working with your phone in a locker on the other side of the gym, and the sets reconcile when the two reconnect.
With your permission, Apple Health sends heart rate, resting heart rate, heart-rate variability (the beat-to-beat change in your heart rhythm), active energy, sleep, breathing rate, and — on Apple Watch models with the sensor — your wrist temperature during sleep. It also reads menstrual flow and menopausal state, if you or another app record them there, and never writes cycle data back. Live heart rate and calories from a Watch workout flow into the workout summary, and Rhythm writes your finished workouts and active energy back to Health.
Confidence, not load. Clear and consistent signals mean today's suggestion is offered with more of it; thin signals, or two devices that disagree, mean Rhythm says so instead of guessing. Your working weights come from the sets you have actually logged.
A raw number left on its own. Rhythm turns what Health sends into plain words and says what it means for today, and every reading in the app opens to a “why” in one tap.
Wrist temperature never sets your cycle phase. At most it changes how a cycle estimate is worded, and only with hedged, medium confidence. Your cycle dates and phase are worked out on your phone and are never sent to our servers, whatever you connect. And a wearable is optional: Rhythm reads your training log first.
Yes: connect Oura in Settings and Rhythm reads your recovery signals and overnight temperature trend from it to adjust how confident today's suggestion is, never the weight you are asked to lift.
Once connected, Rhythm reads a daily summary: your recovery signals, your sleep, your heart rate and heart-rate variability (the beat-to-beat change in your heart rhythm), and how far your overnight temperature sat from your own recent baseline. It reads nothing else.
Confidence, not load. When the signals are clear and consistent, today's suggestion is offered with more of it; when they are thin, or two devices disagree, Rhythm says so instead of guessing. Your working weights come from the sets you have actually logged.
Rhythm never shows Oura's own scores back to you. What you get is Rhythm's plain-language read on your day, not a number to interpret.
Rhythm does not pin your ovulation on a temperature reading. That is a deliberate choice, not a missing feature — the app used to do it and we took it out. One night is never enough, no study has put a finger sensor and a wrist sensor on the same person across the same cycles, and the validated research is about cycles where ovulation happens. If you are on hormonal contraception, have PCOS, are in perimenopause, or your cycles are irregular, a temperature trend is much weaker evidence. Your phase in Rhythm comes from the dates you enter, is worked out on your phone, and is never sent to our servers.
For anything about getting pregnant or about contraception, talk to a doctor, midwife or nurse. Rhythm is a training log, not a fertility tool, and nothing here is medical advice.
Yes: connect WHOOP in Settings and Rhythm reads your recovery signals from it to adjust how confident today's suggestion is, never the weight you are asked to lift.
WHOOP connections are in an early-access window today. A limited number of seats are open, so a connection can be refused if the window is full when you try to connect, and more seats open as the integration grows — try again if that happens.
Once connected, Rhythm reads a daily summary: heart-rate variability (the beat-to-beat change in your heart rhythm) in milliseconds, resting heart rate, sleep, breathing rate, workout calories, and your daily and per-session activity-intensity values. The connection tokens are exchanged and stored on our servers, encrypted, never in the app bundle; disconnecting removes them along with the stored daily snapshots.
Confidence, not load. Clear and consistent signals mean today's suggestion is offered with more of it; thin signals, or two devices that disagree, mean Rhythm says so instead of guessing. Your working weights come from the sets you have actually logged, and nothing a wrist strap sends changes them.
Rhythm never displays WHOOP's own recovery, exertion or sleep-quality scores. Those stay in the WHOOP app. What Rhythm shows you is its own plain-language read on your day, built from the raw measurements, not WHOOP's scoring.
WHOOP's overnight temperature reading does reach Rhythm, but it never sets your phase. At most it changes how a cycle estimate is worded, and only with hedged, medium confidence, because a wrist sensor has no validation study behind it for this. Your cycle dates and phase are worked out on your phone and are never sent to our servers, whatever you connect. And a wearable is optional: Rhythm reads your training log first, so it works with nothing connected at all.
Yes, through Apple Health: any wearable whose app writes its data to Apple Health can feed Rhythm, because Rhythm reads from Apple Health, not from the wearable.
Rhythm reads from Apple Health, not from any wearable directly, so what shows up depends on what you have shared with Health. The Apple Watch and Apple Health page has the full list of what Rhythm reads and what it writes back.
What reaches Rhythm depends on what your wearable's own app chooses to write. Garmin Connect, for example, can write your workouts, heart rate and sleep to Apple Health once you turn that on — but a score like Body Battery is Garmin's own calculation, not an Apple Health data type, so it never comes through, however you set the toggles.
Open your wearable's app, find its Apple Health setting (sometimes called “Health sharing”), and turn on the data types you want it to share. Then, in Rhythm, connect Apple Health the same way you would for an Apple Watch.
There is no direct connection between Rhythm and Garmin, or any other wearable brand — Apple Health is always the bridge. None of this is required: your logged sets always decide what Rhythm suggests; a connected wearable only changes how confident a suggestion is.
It changes how confident today's suggestion is, and it never sets the weight; your logged performance always does that.
Here is the whole rule in one line. Your logged performance drives the weight. Wearables move confidence. A symptom you log can ease a session. Your cycle phase gives context. Nothing on that list except your own logged sets goes anywhere near the barbell.
A recovery reading is a measurement of one night, not a verdict on a person — and a wrong verdict is not harmless. Being told your sleep was poor can make the session worse even when the sleep was fine. So Rhythm uses those signals to say how sure it is, and leaves the number on the bar to the thing it can actually see: what you lifted last time.
Every read in the app opens. Tap the day's suggestion and you get the reasons behind it in plain words. Tap the confidence line and you get what fed it. Tap the “adjusted today” pill on an exercise and you get what changed and why. There is no number in Rhythm you cannot open.
When the data is thin, Rhythm says so rather than filling the gap. A low reading is not a diagnosis, and nothing the app tells you about your recovery is medical advice. And when it suggests an easier day, it names something you can still do — the door stays open.
Training
Rhythm watches your own log for two things: a lift flat at the same weight for several sessions, and four weeks of sustained high training volume.
The first signal is a lift that has stopped moving. Rhythm works out an estimated one-rep max for each lift from the weight and reps you record, and when three sessions in a row land inside the same narrow band — within about 2.5% of each other — it calls that flat. It waits until you have at least twelve logged sessions before it uses the word at all, because three sessions out of five is noise, not a pattern.
The second is volume that has stayed high for a month. Rhythm adds up your weekly training volume, and if your average over the last four weeks sits at 90% or more of your biggest week in the last eight, that is a sustained stretch. It needs at least eight logged workouts before it can tell.
When either one fires, the suggestion has the same shape: about 10% off your working weight, or a set or two fewer, for a week.
Coaches watch a third thing — effort creeping up at the same weight, where the set that used to leave you two reps short now leaves you none. Rhythm does not track it. Logging reps in reserve is not in the app today, so plateaus are read from the weight and reps you record instead. We would rather tell you that than let the picture look complete when it is not.
No study fixes three sessions, or 2.5%, or 90%. We looked, and found no peer-reviewed rule for spotting a stall from logged sets alone. Those thresholds are our judgement, set slow enough to ride out normal week-to-week noise, and what would prove them wrong is simple: if they call a deload for people who did not need one, or stay quiet while people stall, they are wrong and we will change them.
An easier week is still a training week. A walk, some mobility, a lighter session on the same lift — all of it counts.
Honestly, the evidence is mixed: one controlled study found a mid-programme deload did not add strength and cost some in the lower body, so Rhythm treats a lighter week as easing fatigue, not a guaranteed boost.
The study is Coleman and colleagues, 2024, in the journal PeerJ. Thirty-nine trained lifters ran a nine-week programme; half stopped lifting for a week at the halfway point and half trained straight through. Muscle size, power and muscular endurance came out much the same in both groups. Lower-body strength did not — the group that trained through gained more.
Read that carefully, because it is a week off, not a lighter week, and it is one study with a small group. It does not prove a deload is bad. What it does is remove the comfortable idea that a deload reliably buys you a jump afterwards.
Coaches deload, and when asked why, they mostly say it is to reduce fatigue rather than to add performance (survey, 2024). Asked how they decide, they describe reading the athlete rather than hitting a threshold, usually every four to six weeks (interview study, 2023). The physiology underneath is less settled than it sounds: a 2024 practical review of deloading states that the evidence for fatigue accumulating across resistance training is indistinct, and that what builds up may be muscle damage instead (Bell, 2024).
It offers a lighter week as a low-cost option when your log shows a lift flat or a month of high volume, and it says what that is: an easier stretch, not a promised jump. Rhythm does not tell you that fatigue has built up and a lighter week will clear it, because nobody has shown that cleanly.
And a lighter week stays a training week. Lighter sets, a walk, some mobility — the door does not close.
When your estimated max has been flat for a few sessions, Rhythm suggests a short deload or a variation of the lift, a common next step and not a promise.
Flat has a definition in the app. Rhythm tracks an estimated one-rep max for each lift from the weight and reps you log, and when three sessions in a row land inside the same narrow band, it flags that lift. It stays quiet until you have around twelve sessions of history, so a slow start is never mistaken for a wall.
A short lighter week, or a variation of the same movement. A different rep range. A pause at the bottom of each rep. These are the things lifters and coaches usually try next when a lift stops moving — that is the whole claim, and it is not a promise. Whether any of them restarts your progress depends on you, and we cannot know that in advance.
It will not add weight to break a stall. Telling you to push harder on a number nobody worked out is how a training app gets someone hurt, so when a lift is flat, the suggestions move down or sideways, never up. Weight goes up when your logged sets say it can, and that happens on its own once the reps come back.
A lighter week is not a week off. Keep showing up, keep the movement in, and let the numbers come back to you.
RIR is how many more reps you had left in the tank at the end of a set.
Two reps in reserve means you stopped with two more in you. Zero means you could not have done another one. Coaches use it because the same weight is a different job on different days: 100 pounds with four reps left is easy work, and the same 100 pounds with nothing left is close to your limit. RIR describes the effort, not the load.
Reasonably well close to failure, and less well the further away they are. The scale comes from Zourdos and colleagues in 2016, and the research that followed finds trained lifters estimate it most accurately on hard sets and get vaguer on easy ones (review). So it is a useful number when the set was hard and a noisy one when it was not — worth knowing before you build a whole programme on it.
Rhythm does not ask you for RIR today. It reads plateaus from the weight and the reps you actually logged, which are facts rather than estimates.
The cost of that choice is real: Rhythm misses the sign where effort climbs while the weight stays the same. That is why the app watches two signals rather than three, and why we say so on the page about deload signs rather than letting you assume it sees everything.
No: an easier day in Rhythm always names something you can still do, a walk, a stretch, a lighter session, and the door is never closed.
An easier day turns up when your own log or your own signals point that way — a month of high volume, a lift that has gone flat, a symptom you recorded. It is not a daily habit and it is not a mood. And when it arrives, it arrives with options.
An easier day is an easier day. It is not a closed door. Lowering the load is caution; taking every option off the table is not caution, it is discouragement, and it costs the thing the app exists for — that you keep showing up. So every easier-day message in Rhythm names something: a walk, a stretch, some mobility, a lighter session, a muscle group you rested yesterday. “Rest today” on its own is not an answer the app is allowed to give.
That is not a good intention, it is a test. A check in our build fails if any of those messages stops offering a way to move, so the rule cannot quietly rot.
When the signals genuinely point to illness or injury, or on a screen written against medical guidance — the first weeks after giving birth is the clearest case — the honest answer can be rest, and Rhythm gives it plainly. Even there it names what is safe to do rather than stopping at what is not.
Every suggestion is a suggestion. Open the routine you planned and train it. Rhythm logs what you did and learns from it, and that is the part that decides what it says tomorrow.
Your cycle
A rise in your overnight temperature held over several nights, not one, is the honest signal, and a device can only show it after it happened.
Your resting body temperature sits a little higher after ovulation and stays there. That is why one warm night tells you nothing on its own: the signal people actually use is a small rise that holds for several nights in a row, read against your own normal rather than a chart built from other people.
Because the rise comes after the event, a ring or a watch is looking backwards. It can say that has probably already happened. It cannot say that is about to happen this week. Oura publishes its own peer-reviewed validation of its temperature-based ovulation detection (JMIR, 2025), and Apple states that the ovulation estimates its Watch produces are retrospective. We have not tested either device ourselves, so we do not put a number of our own on how well they work.
Every validated study we could find looked at cycles where ovulation happened. If you are on hormonal contraception, or you have PCOS, or you are in perimenopause, or your cycles are irregular, a temperature pattern is much weaker evidence, and sometimes it is no evidence at all. We would rather say that plainly than let an app imply a certainty it does not have.
Rhythm never tells you that you ovulated. It does not confirm ovulation from a device reading, and a temperature shift does not move your phase dates. Your phase is worked out from the dates you enter, on your phone, and those dates are never sent to our servers.
If you are making a decision about getting pregnant or about contraception, talk to a doctor, midwife or nurse. Rhythm is a training log. It is not a fertility tool, and nothing here is medical advice.
One that learns your own pattern from what you log, instead of applying a table built from other people.
Bodies do not agree with each other. A signal that means back off for one person barely registers for the next, and cycle phase is a clear example: the average difference across a group is smaller than one person's own swing from day to day. A table that says do less this week because of the date is therefore wrong for a large number of the people it is applied to, and there is no way to tell in advance which ones.
Your working weights come from the sets you have logged, not from a phase. Your recovery is measured against your own baseline rather than an average person's, and in the luteal phase against your own history from that same part of your cycle. If you log a symptom — cramps, low energy, poor sleep — the day's suggestion can ease off. It can never harden. A symptom you report is allowed to make a day gentler and is never allowed to make it heavier.
It rules out telling you what your body is doing when we cannot know. Rhythm suggests; it does not diagnose, and it does not demand. And an easier day is never an empty one — there is always something named that you can still do.
Cycle tracking is optional and open to anyone, whatever your gender. Your cycle dates and phase are worked out on your phone and never sent to our servers.