Apple Watch Series 12 Health Features: Readiness, Recovery HRV, Health Age, Quest Labs and What Changed in HealthKit

At WWDC in June, Apple delayed the health feature everyone expected. Project Mulberry, the AI coach, and the Apple Health+ subscription that was supposed to carry it stayed off the keynote so Apple could rebuild the Watch’s heart-rate engine first. On 9 September it shipped that engine and started interpreting its output: a readiness score from 0 to 10, a Health Age, an Insights tab and a Quest lab panel inside the Health app [1][2].

The coverage since has focused on the score. The more useful story for anyone building on Apple health data is the asymmetry underneath it. Apple opened more raw signal than it opened interpretation. HealthKit gained a new heart rate variability type, structured workout zones and two menopause types. Readiness, Health Age and Insights gained no type at all. Apple is giving developers better inputs and keeping the outputs for itself.

What did Apple ship for health on 9 September?

Short answer: new sensing and a readiness score on the two newest watches, an interpretation layer in a Health app that has not shipped yet, and a handful of genuinely new HealthKit types. The way to read the announcement is not the feature list. It is the gate next to each feature [1][2][5][6].

GymKit heart-rate sharing through AirPods Pro 3, Workout Buddy working without a nearby iPhone, and better treadmill distance also landed [1]. They are incremental, and the iOS 27 beta playbook already covers the catch on each.

Readiness: a category default, with a seven-night cold start

Short answer: Apple turned a specialist product into a Watch default. That legitimises the category. It does not hand you the score.

Apple’s readiness score runs from 0 to 10 and comes with one of four plain-language recommendations: Recover, Pace Yourself, Ready or Go For It. Apple’s Watch release describes the inputs as recent activity, training load, vitals and sleep score, and reviewers list the vitals as resting heart rate, HRV, respiratory rate, sleep duration and blood oxygen [2][18]. It updates through the day as workouts and daytime vitals arrive, which Oura’s morning score does not [1][9]. Apple says the algorithm was developed with data from the Apple Heart and Movement Study [1]. That is a construction claim, not independent validation of the composite. How Apple’s three inputs compare with Oura’s nine contributors, WHOOP Recovery and Garmin’s two numbers is in the readiness score map.

The detail most coverage skipped is the cold start. The score needs at least seven nights of data before it appears, and most reviewers had not seen one after a week of testing [8][9]. 9to5Mac’s single comparison, an 8 on Apple against an 85 on Oura the same morning, is an anecdote, not a calibration [8]. Apple has not said which integers sit under which label.

For product teams, four consequences follow.

Consumers will expect the shape. A single daily number with a verb is now what a new Apple Watch is. Matching that on Apple-only, for the subset of users who just bought a Series 12, is table stakes on one SKU, not a strategy.

Apple’s score is not yours. It is not a HealthKit type. You cannot query it, re-weight it, or show it to an Android user or to someone on a Series 10. A readiness signal that degrades to phone data when no wearable is present is a different product problem from decorating Apple’s 0 to 10.

The first week is still yours. Apple’s own score is silent for seven nights. Every new user of a health product faces the same gap, and the cold-start problem is solved with population priors and phone data, not by waiting.

Legitimisation cuts both ways. Apple has told the market that daily readiness is a default. That helps anyone already selling the idea and hurts anyone whose only differentiator was having a number Apple did not.

The Health Sensing System, and why the richest data still needs a new watch

Short answer: the sensing upgrade is how Apple sells Series 12 and Ultra 4. Health is the attach, not the coverage strategy.

Series 12 and Ultra 4 have new optical and electrical heart sensors with larger, more power-efficient green LEDs. Heart rate is measured every five seconds whether the wearer is moving or still, against roughly every five minutes at rest before. HRV is measured as often as every five minutes at rest, which Apple puts at 24 times more often than before [1][2][8]. Apple published a study of more than 1,000 participants claiming the highest heart-rate accuracy among leading wearables available as of June 2026 [2][3]. Battery life holds at up to 24 hours on Series 12 and rises to 50 hours on Ultra 4 [2][8][15].

Hypertension notifications are not the September headline. They are a 2025 feature that reads 30-day optical patterns on Series 9 and later, not a cuff reading. Sleep apnea notifications continue on Series 9 and later, Ultra 2 and later, and SE 3 [2].

None of the new sensing reaches the installed base. watchOS 27 installs on Series 9 and later, SE 3 and Ultra 2 and later, but readiness, five-second heart rate and Recovery HRV do not come with it [1][17]. Health Age requires a Watch. Most adults still do not wear one: smartphone ownership runs far ahead of wearables, and the units that produce rich health data are concentrating in the premium half. Everyone outside the newest-Watch slice gets a redesigned Health app later this year, not the new sensing.

For developers, denser samples are another overlapping source. Series 12 can measure up to 17,280 heart-rate readings a day, although Apple describes measurement, not how much of it lands in HealthKit [11]. Apps that mirror raw samples into their own storage should aggregate per minute or per five minutes and move to anchored queries, and the new stream sits alongside phone motion, GymKit and third-party wearables in exactly the deduplication problem that gets worse every time a sensor improves.

What actually changed in HealthKit

Short answer: more than the keynote suggested, and all of it is inputs. The developer documentation for iOS 27 adds three things most health apps will touch [10][11][12].

A second HRV type. heartRateVariabilityRMSSD now sits next to heartRateVariabilitySDNN [12]. This matters more than any keynote slide. SDNN, the spread of beat-to-beat intervals over a window, is what Apple Watch has always reported. RMSSD, built from successive differences, is the measure Oura, WHOOP and most recovery research use, and it is more stable over short overnight windows. The two are not convertible. Apple has not said explicitly that Recovery HRV is RMSSD and overall HRV is SDNN; reviewers and developers infer it from the documentation, and it should be verified on real Series 12 data before anyone builds on it [13]. What is certain is the practical consequence: a user’s HRV history now comes in two incompatible series, each needing its own baseline, and any “what is a good HRV” comparison has to know which one it is looking at.

Workout zones as structured data. HKWorkoutZoneConfiguration, HKWorkoutZoneGroup and a preferredWorkoutZoneConfiguration(for:) method let apps read a person’s zones, whether they set them or the system generated them, and read time in zone from completed and live workouts [10]. Every training app has maintained its own zone model. Apple now keeps one.

Menopause as data. Two new category types, menopausalState (a point-in-time sample: menopause, perimenopause or none) and bleedingAfterMenopause (an interval with an intensity, clinically distinct from menstrual flow), are readable and writable by third-party apps [10][14]. The cross-platform libraries picked them up within two days of the event [14].

What did not change. There is no readiness type, no Health Age type and no Insights feed. Heart rate at five-second cadence stays on the existing heartRate type [11]. HealthKit still measures and does not interpret, still does not span Android, and still does not cover users without the hardware. The asymmetry is the story: Apple made the raw material richer and kept every derived score inside its own apps.

Did Apple launch an AI health coach?

Short answer: no. Mulberry arrived as Insights, inside the free Health app, later this year, with no Health+ subscription.

The Insights tab surfaces a summary that updates through the day across heart, sleep, readiness, fitness, vitals and cycle tracking, plus a For You section of personalised recommendations and articles, such as a suggested wind-down routine after a stretch of late nights [1][4][5]. That is the coach, scaled back and unnamed. Reporting before the event said the fuller Mulberry service and a standalone Health+ offering had been folded into the ordinary Health app [7]. September confirmed the fold. As of the first reviews, the redesigned app was available only in beta [8][16].

That is a pricing decision, not a technical one. Two days before the event we mapped what consumers pay for health insight: Oura at $5.99 a month, Garmin Connect+ at $6.99, Google Health Premium with its Gemini coach at $9.99. Apple’s Health app was on that map at $0 and still is. Google shipped a paid coach in May. Apple had the distribution, the sensor and three months after WWDC, and still chose to sell watches rather than a health subscription.

The consequence for everyone in the $6 to $10 band is real pressure on what an interpretation tier can charge on iOS. Plain question-and-answer over health data was already free in ChatGPT; a daily summary is now about to be free in the Health app. What stays sellable is what Apple’s version lacks: coverage beyond the newest Watch, Android, and a view across sources Apple does not control. That is why the AI coach race did not end on 9 September.

What do Quest labs inside the Health app change?

Short answer: Apple can now hold daily optical data and a yearly A1c in one place. Third-party apps still cannot read the fusion.

Once the redesigned app ships, users will be able to book a panel of more than 50 biomarkers for $119 at about 2,000 Quest Diagnostics locations in the US and see the results in Health alongside Watch, iPhone, AirPods, clinical records and third-party app data [1][4]. Oura already sells a Quest panel for $99 and WHOOP sells Advanced Labs from $199 a year, now without a membership. The consumer blood-testing boom was well underway before Apple made the default Health app a storefront for it.

The strategic shape is the one in what builds a moat in health data: a daily optical sensor is one modality, a blood panel is another, and whoever holds both owns a record nobody can backfill. Labs were Apple’s missing modality. HealthKit’s clinical records API already exposes lab results that arrive from connected health systems, but Apple has not said whether panels bought inside the Health app will be readable the same way, and the interpretation that puts A1c next to VO2 max is not exposed at all.

Health Age and the Longevity tab

Short answer: biological age stopped being a differentiator on iOS.

The Longevity tab scores heart health, sleep, mental wellbeing, movement health, metabolic health, hearing and nutrition. Health Age sits inside it, built from VO2 max, resting heart rate, sleep and HRV, with optional A1c and LDL, and shows how those metrics track against chronological age [1][4]. Movement evaluations use the iPhone camera and Watch to assess flexibility, strength, balance, movement mechanics and VO2 max, on device, with no video recorded or stored [1]. The VO2 max test can also run with AirPods Pro 3 or a third-party heart rate device [1].

Wearable biological age was a specialist category in March. It is now a tab in Apple Health. Shipping “we have a bio-age number” is no longer a reason a user stays in an iOS app. What remains differentiating is whether the number works without a Watch, whether its factors are explainable, and whether anyone has validated the composite rather than citing its components. Apple says the algorithm is “informed by the best available science” [1]. That describes how it was built, not how well it predicts.

Perimenopause becomes a platform dimension

Short answer: life stage is now first-class Health data, readable by apps. Recovery models that ignore it will look dated next to the operating system.

Users can log perimenopause or menopause and their symptoms, and get a prompt to speak with a doctor if they log bleeding after menopause. Users 40 and over can get a notification when their logged cycle patterns suggest perimenopause [1]. Unlike most of the event, this reached developers: the two new category types mean an app can read a user’s recorded menopausal state and respond to it [10][14].

The durable point is the one in women’s health data: resting heart rate, HRV and temperature shift across the cycle and through the menopausal transition in ways most scoring models still ignore. Apple has made the transition a structured field. A product that computes recovery as if every user were a 30-year-old man is now competing with the Health app on the same phone.

What this means for builders

The posture is triage, not panic and not imitation.

1. Adopt the new inputs now. Request heartRateVariabilityRMSSD alongside SDNN, keep them as separate series with separate baselines, and verify on Series 12 data which one Recovery HRV writes. Read workout zones from HealthKit instead of maintaining a parallel model where you can. Handle the menopause types.
2. Plan storage for dense heart rate. Aggregate on ingest and use anchored queries. Five-second sampling is an optional upgrade for a minority of users, not a new baseline.
3. Do not build on Apple’s scores. Readiness, Health Age and Insights are not data sources. Treat a keynote screenshot as a feature request from users, not as an input.
4. Do not feature-match on Apple only. Those features exist on the newest Watch in US English. Design for mixed hardware, Android and the people who will never wear a Watch.
5. Price for the free default. If your paid tier is a daily summary, Apple is about to give one away on iPhone. Charge for coverage, continuity and a view across sources.

When Apple’s sensor improves, the signal should flow straight through your pipeline. When Apple ships a score, that score should not become your product’s load-bearing input. That is the problem a cross-platform data layer exists to solve, and it is the one we work on at Sahha.

The short version

Apple Watch Series 12 ($399) and Ultra 4 ($799) measure heart rate every five seconds and HRV as often as every five minutes, add Recovery HRV, and show a 0 to 10 readiness score that needs seven nights of data and does not come to older watches. The redesigned Health app, still in beta, adds a free Insights tab instead of a paid coach, Health Age, movement evaluations and a $119 Quest panel. HealthKit gained a new RMSSD heart rate variability type, workout zones and two menopause types, and gained no type for readiness, Health Age or Insights. Apple opened the inputs and kept the outputs.