Apple spent years building an AI health coach. Then it killed the project before anyone got to use it.

In early February 2026, Apple quietly wound down Project Mulberry, an internal initiative to build an AI-powered health coaching service. The project had physicians, sleep experts, nutritionists, and mental health professionals training AI models. There were plans for educational video content. The whole thing was supposed to be Apple’s answer to the growing AI fitness coaching market.

It never shipped. Eddy Cue, Apple’s Services chief, took over the health division and concluded that competitors like Oura and Whoop already offer more compelling features. Some of the planned capabilities will trickle into the existing Health app over time. But the standalone AI coach? Dead.

If you were hoping Apple would finally make the Health app useful for actual training guidance, you need a new plan. Here’s what’s available right now, what each option does well, and which one fits your situation.

What Apple Was Building (and Why It Mattered)

Project Mulberry wasn’t just another chatbot slapped onto a health app.

The original vision involved AI trained on data from Apple-hired physicians, sleep experts, nutritionists, and mental health professionals. Think of it as a virtual health team that understood your Apple Watch data and could give personalized recommendations based on clinical expertise, not just generic “move more” nudges.

The educational video content component was supposed to make it feel less like a data dashboard and more like having a coach walk you through sleep hygiene, recovery protocols, and nutrition basics. That’s a meaningful gap in what Apple Health currently offers. The app tracks everything but tells you almost nothing about what to do with the information.

Why Cue pulled the plug makes sense from a business angle. Oura, Whoop, Fitbod, and others have spent years building exactly these features. Apple entering the market with v1.0 would’ve been competing against products with millions of users and years of iteration. Instead, they’ll fold select features into the existing Health app incrementally. Translation: you might get slightly better sleep insights on your Apple Watch in 2027. Maybe.

For anyone who trains seriously, that timeline is useless. You need coaching tools that work today.

The Alternatives That Fill the Gap

Project Mulberry promised four things: AI coaching, sleep optimization, nutrition guidance, and mental wellness support. No single app does all four well. But a combination of two or three covers the territory Apple was aiming for.

Here’s how the current options break down by what Mulberry was supposed to deliver.

For AI Training Coaching: Fitbod

Price: $79.99/year What it does: Generates workouts using machine learning trained on 100+ million logged sessions. Adjusts for recovery, available equipment, and muscle group fatigue.

Fitbod is the closest thing to what Apple’s AI coach would have looked like for strength training. The algorithm doesn’t just increment your reps by one each week. It tracks fatigue patterns across muscle groups and programs deload weeks when your performance data suggests accumulated fatigue. We tested four AI fitness coaches head-to-head over 12 weeks, and Fitbod produced measurably better strength gains than the alternatives.

The limitation: Fitbod is strength-focused. If you run, cycle, or do mixed training, it won’t program your cardio. You’ll need something else for endurance work.

Best for: Lifters at any level who want programming that actually adapts. Beginners who need structure. Intermediate lifters who’ve plateaued on cookie-cutter programs.

For Sleep and Recovery: Oura Ring 4

Price: $349 for the ring + $5.99/month for a membership What it does: Passive sleep tracking (sleep stages, HRV, skin temperature, respiratory rate), readiness scores, activity tracking. The redesigned app launched earlier this year with improved sleep insights.

This is the area where Cue explicitly said competitors had Apple beat. Oura’s sleep tracking is more granular than what the Apple Watch provides, and the readiness score gives you actionable morning data: train hard today, go easy, or take a rest day. The ring form factor means you actually sleep with it on, unlike a chunky watch.

The February 2026 app redesign added trend analysis that connects sleep patterns to daytime energy and workout performance over weeks, not just single nights. That’s the kind of longitudinal insight Apple was trying to build into Mulberry.

The limitation: Oura doesn’t coach you. It tells you what happened and gives a readiness number. Interpreting that data and adjusting your training is still on you.

Best for: Anyone who suspects their recovery is the bottleneck. People who train consistently but feel crushed some days and great on others without understanding why.

For Recovery Monitoring: Whoop 5.0

Price: $199/year (basic) to $359/year (premium) What it does: 24/7 HRV, strain tracking, sleep monitoring, recovery scores. The Whoop 5.0 and Whoop MG shipped with improved sensors and the new Whoop MG added ECG.

Whoop and Oura overlap significantly, but Whoop is better for people who train hard and want strain quantification. The strain score tells you how much physiological stress your workout actually produced, regardless of what your program said you were supposed to do. A “light” session that spiked your heart rate gets captured accurately.

The coaching angle comes from Whoop’s recovery recommendations. Green means go. Red means back off. Yellow means adjust intensity. It’s blunt, but for athletes who tend to overtrain, that binary signal prevents a lot of damage.

The limitation: It’s a subscription device. Stop paying, stop getting data. At $199-359/year, you’re committing to ongoing cost. And Whoop doesn’t program workouts. It just tells you how much capacity you have.

Best for: High-volume trainers, endurance athletes, anyone who chronically overtrains. CrossFit athletes and triathletes get the most value from strain tracking.

For Nutrition Guidance: MacroFactor

Price: $71.99/year What it does: Calorie and macro tracking with an expenditure algorithm that adjusts your targets based on actual weight trends, not static TDEE calculators.

Apple’s nutrition coaching component was supposed to use AI trained by nutritionists. MacroFactor doesn’t use AI in the marketing-hype sense, but its expenditure algorithm is genuinely smart. It watches your weight trend over time and recalculates your actual energy expenditure weekly. If you’re eating 2,400 calories and maintaining weight, it knows your TDEE is around 2,400, regardless of what a formula predicted.

The MacroFactor Workouts feature that launched in January 2026 added auto-progressing strength programming to the nutrition app, making it a dual-purpose tool.

The limitation: It won’t tell you what to eat. It tracks what you eat and adjusts targets. If you need meal plans or recipe suggestions, you’ll need to look elsewhere.

Best for: People whose main goal is body composition change (fat loss or muscle gain) and who are willing to log food consistently.

For AI Running Coaching: Runna or URUNN

Price: Runna $14.99/month; URUNN free tier available What it does: AI-generated running plans that adapt to your fitness level, race goals, and feedback.

If your training is primarily running, Apple’s AI coach would’ve been competing with dedicated AI running coaches that have been iterating for years. Runna builds periodized training plans for specific race distances and adjusts week to week. URUNN offers a free tier that’s surprisingly functional for casual runners.

The limitation: Running-only. If you do mixed training, these won’t help with your strength or cross-training days.

Best for: Runners training for a specific race who want structured plans that adjust to their progress.

What About Apple Health App Updates?

Apple said some Mulberry features will roll into the existing Health app incrementally. Based on Apple’s track record with health features, here’s what that likely means.

Realistic timeline: Don’t expect anything before WWDC 2026 at the earliest, with features shipping in iOS 20 (fall 2026) or later. “Incrementally” is Apple-speak for “spread across multiple OS releases over two or more years.”

What will probably ship first: Enhanced sleep analysis is the safest bet. Apple Watch already collects the raw data. Improved sleep stage detection and maybe a readiness-style score could arrive without new hardware.

What will ship last (if ever): The nutrition coaching and mental health components. These require partnerships, liability considerations, and regulatory navigation that Apple moves slowly on. The FDA’s evolving rules on fitness wearables add another layer of complexity.

The honest assessment: If you’re waiting for Apple to deliver Mulberry’s vision through Health app updates, you’re waiting 2-3 years for features that exist in other apps today.

How to Build Your Own AI Health Stack

Since no single app replaces everything Mulberry promised, here are three practical stacks based on budget and training type.

Budget Stack ($72/year)

• MacroFactor ($71.99/year) for nutrition tracking and basic workout programming
• Apple Health (free) for sleep data from your Apple Watch
• Total: ~$72/year

This works if you’re primarily focused on body composition and don’t need advanced recovery metrics. Apple Watch sleep tracking isn’t as detailed as Oura, but it’s decent for basic sleep hygiene awareness.

Mid-Range Stack (~$150/year)

• Fitbod ($79.99/year) for AI-driven strength training
• MacroFactor ($71.99/year) for nutrition
• Apple Health (free) for sleep
• Total: ~$152/year

The best balance of coaching quality and cost. Fitbod handles your training programming, MacroFactor handles nutrition, and your Apple Watch covers basic sleep tracking. This is close to what Mulberry was supposed to deliver, minus the educational video content.

All-In Stack (~$500/year)

• Fitbod ($79.99/year) for strength programming
• Oura Ring 4 ($349 one-time + $71.88/year membership) for sleep and recovery
• MacroFactor ($71.99/year) for nutrition
• Total: ~$573 year one, ~$224/year after

This is overkill for most people. But if you train 5-6 days a week and want data-driven decisions for training, recovery, and nutrition, it covers every angle Apple was targeting. The Oura Ring handles what Apple Watch can’t: granular sleep staging, skin temperature trends, and a readiness score that actually influences your training decisions.

Who Wins From Apple’s Retreat

Oura and Whoop are the clearest winners. Cue specifically cited them as reasons to kill the project, which means Apple’s own leadership validated their products. That’s free marketing you can’t buy.

Fitbod and MacroFactor win by default in the coaching space. Apple never shipped a competitor, so they keep growing without a 2-billion-device ecosystem breathing down their neck.

The losers are Apple Watch users who wanted a smarter Health app. You’ll get incremental improvements. But the vision of a unified AI health coach that ties your watch data to personalized training, sleep, and nutrition advice? That died with Project Mulberry.

The good news: the alternatives that exist today are better than what Apple’s v1.0 would’ve been. These companies have years of real user data and iteration behind them. Apple would’ve been starting from scratch, no matter how many physicians they hired.

Pick the stack that fits your goals, budget, and training style. Don’t wait for Apple.

---

Analysis based on Project Mulberry reporting from Bloomberg and hands-on testing of recommended alternatives. Prices current as of March 2026.