At its Worldwide Developers Conference in June 2026, Apple shipped a redesigned Health app and improved sleep tracking in watchOS 27 — but conspicuously held back the AI health coach many expected, internally known as Project Mulberry. According to reporting from Sahha and coverage of WWDC 2026, Apple pushed the coach, along with a planned Apple Health+ subscription, past the developer conference entirely.
The delay is notable given how long Apple has telegraphed health ambitions. Apple Watch already carries FDA clearances for its ECG and irregular-rhythm notification features, clearances that took years of engineering and regulatory work to secure. That history appears to be shaping Apple’s more conservative posture toward an AI coach: unlike a fitness-tracking feature that can afford to be occasionally wrong, a coach dispensing health advice sits closer to the regulatory line Apple has already had to navigate carefully once before.
Apple’s caution also reflects lessons from its own past stumbles: earlier health-adjacent features, including blood-oxygen sensing, drew patent litigation that forced Apple to disable the feature on newer U.S. watch models sold after a partial import ban. That experience appears to have reinforced an institutional wariness about shipping health features before both the underlying technology and its legal footing are fully settled, a posture that stands in visible contrast to the faster, more iterative approach Google and the wearable-specialist companies have taken with their own AI coaching tools.
What Apple actually shipped at WWDC
iOS 27 introduced a redesigned Health app with new perimenopause and menopause tracking, while watchOS 27 rebuilt the underlying heart-rate tracking engine, refined sleep-stage detection, and upgraded the Workout Buddy feature with more granular pace, distance, and duration feedback drawn from a user’s fitness history. Siri also gained broader Apple Intelligence integration system-wide. None of it, notably, included the conversational AI health coach that had been rumored for over a year.
Why Apple held back
Apple’s stated reasoning, as described in coverage of the announcement, is that it “refocused near-term engineering on strengthening the reliability of the underlying biometric data” before layering an intelligence layer on top. The logic: an AI coach is only as trustworthy as the sensor readings it reasons over, and a confident recommendation built on a noisy heart-rate sample or a misclassified sleep stage doesn’t just fail to help — it actively erodes user trust in the whole system. That is a notably more cautious posture than rivals have taken.
The competitive contrast
The delay stands out because competitors moved in the opposite direction throughout 2026. Google shipped a Gemini-powered “Health Coach” as part of its Fitbit-to-Google Health rebrand in May 2026, priced at $9.99 a month, while Whoop, Oura, and Samsung all rolled out proactive AI coaching features across the same stretch of months. Apple, by contrast, is reportedly still planning an Apple Health+ subscription that would eventually bundle food tracking, nutrition planning, and AI-driven health advice, but has not committed to a firm release date.
The case for caution — and against it
Apple’s caution has real merit: healthcare-adjacent AI advice built on faulty inputs is a documented failure mode across the wearable industry, and Apple’s approach of hardening the sensor stack first is a defensible response to that risk, especially given the company’s broader liability exposure with hundreds of millions of Apple Watch users. The counterargument, made implicitly by rivals racing ahead, is that iterative AI coaching with clear disclaimers can still deliver value even before every underlying signal is perfect — and that Apple’s delay cedes a full product cycle of habit formation and subscription revenue to Google, Samsung, and specialist wearable makers.
What’s next
Apple has not announced a new target date for Project Mulberry or Apple Health+, and the company’s pattern of holding major health features until they clear an internal reliability bar suggests it could slip into 2027. In the meantime, the delay leaves an opening: whichever AI health coach establishes itself as the default this year — Google’s Gemini-based assistant, Samsung’s Vitals-driven guidance, or a wearable specialist like Whoop or Oura — may be difficult for Apple to dislodge once it finally ships, even if Apple’s version arrives with more rigorously validated data underneath it. For the roughly 100 million active Apple Watch wearers worldwide, the immediate practical impact of the delay is minimal — watchOS 27’s sleep and heart-rate improvements still ship on schedule — but the strategic stakes are real, since consumer habits formed around a rival’s AI coach this year may prove sticky by the time Apple’s own version finally arrives.
