Apple Fitness+
Fitness+ is built for passive consumption — follow a video, close your rings. This is a pitch for what comes next: expert programming that is dynamic, not static. Apple has Watch biometrics, HealthKit, Fitness+ distribution, and creator relationships. The missing layer is the adaptive AI that connects them.
Prototype
Try the adaptive workout loop.
This is a simulated Push Day session from Jeff Nippard's Upper Strength program. The AI predicts your working weight, you log your set, rate your effort (RPE), and the system adapts the next set in real time. Click Start Workout to begin.
Problem Statement
Expert programming is static. Apple has the data to make it dynamic.
The science-based fitness space has excellent programming — built on RPE periodization, progressive overload, and deload scheduling. Jeff Nippard's Bodybuilding Transformation System is a prime example. The programming is genuinely rigorous. The delivery mechanism is an Excel spreadsheet.
“Manipulating/tracking/viewing an Excel spreadsheet on your phone is tedious, not smart phone friendly, and just plain silly. Such a shame. I went right back to Layne's Workout Builder which, while it's not an app, translates beautifully to a smart phone.”
— Verified buyer, Jeff Nippard program review
Users churn from expert programs not because the programming is wrong, but because the delivery mechanism fails them at the gym door. Meanwhile, Fitness+ serves gym-goers with the same one-size-fits-all video for a 120 lb beginner and a 200 lb intermediate. Apple Watch already knows your HRV, sleep, and heart rate mid-set — and does nothing with it to adapt the session. That's the gap.
Approach
Apple already has all five pieces.
This feature doesn't require Apple to invent anything new. It requires them to connect what already exists. No competitor can do this because no competitor has all five assets simultaneously.
| Asset | What it provides | Competitor equivalent |
|---|---|---|
| Apple Watch | HRV, sleep quality, resting HR, real-time heart rate | Whoop (no content), Garmin (no content) |
| HealthKit | Aggregates all biometric signals into one data layer | Google Fit (no premium content) |
| iOS 26.4 Health app | Native calorie + macro tracking — confirmed for spring 2026 | MyFitnessPal (no hardware) |
| Fitness+ | Content delivery surface with existing creator relationships | Peloton (no Watch), Nike Training (no hardware) |
| Creator network | Science-based trainers with existing programs and audiences | No platform has this combination |
Apple's confirmed Health+ subscription service launching in 2026 — including an AI health coach and nutrition tracking — is the infrastructure layer this feature sits on. The adaptive training pitch isn't speculative. Apple is building the foundation right now.
What I'm Designing
Six capabilities. One adaptive loop.
A new mode inside Fitness+ for gym-goers. Not another follow-along video class. An AI-driven session where creator programming logic adapts to you in real time, every set, every workout, every week.
Jeff provides the periodization architecture — exercise selection, RPE targets per set, deload triggers, plateau thresholds. Not video content. Decision logic. Apple's ML applies that architecture to each user's individual strength curve.
Two layers. Calibration Phase (sessions 1–5): Brzycki 1RM establishes a verified baseline — no day-one overload. Optimization Layer (session 6+): Core ML on-device refines it using population-scale patterns — plateau prediction, fatigue curves, RPE calibration. Data never leaves the device.
RPE feedback after every set drives in-session adjustments. RPE 6 or below → next set increases. RPE 9+ → weight drops. No guessing.
HealthKit aggregates sleep, HRV, steps, and protein intake (iOS 26.4 native tracker) into a daily readiness score that shapes your session intensity.
Week-over-week overload logic, automated deload scheduling every 4th week, and plateau detection surface when your programming needs to evolve.
Creators publish their programs inside Fitness+. Users subscribe at $X/mo. Apple takes 30% year 1, 15% after — same model as Podcasts Subscriptions.
On safety and liability: Apple has navigated this boundary before — ECG, AFib detection, fall detection, and crash detection all ship under the same framework: informational tool, not medical advice. Adaptive Training follows the same model with three built-in guardrails.
- ◆Historical Floor — the system never suggests a weight more than 5–10% above a verified successful lift. Cold-start overload is impossible by design.
- ◆RPE Ceiling — if a user logs RPE 10, the system locks the weight for the next session regardless of what the progression logic says.
- ◆Watch Motion Signal — the accelerometers detect wrist acceleration variance and HR divergence from the expected rep-count curve. A directional flag for unusual effort, not a substitute for a spotter.
The AI recommends. The user loads the bar.
Creator Model
Jeff designs the programming logic. Apple applies it to 100 million users.
The platform shift is what Jeff contributes. Not videos — Apple already has those. Jeff provides the periodization architecture: the conditional logic that determines when load increases, when a deload triggers, what RPE to target at week 4 vs. week 10, and which substitution to suggest when equipment is unavailable. That's years of evidence-based programming expertise encoded as rules. Apple's ML takes those rules and personalizes them to each user's individual strength history.
Jeff's form videos exist inside the app as on-demand reference— not as a class to follow. You tap to check form on an unfamiliar exercise, then get back to your sets. The AI drives the session. Jeff's expertise drives the AI.
Jeff's Bodybuilding Transformation System already has the programming rigor Apple needs: RPE periodization, warm-up protocols, substitution logic, deload scheduling. The program exists. The delivery mechanism is the problem.
At Apple's scale — 100M+ Watch users — a $6.99/mo in-app subscription at even 0.1% adoption means 100,000 subscribers and $700K/mo gross. Jeff's $50 PDF sold to his existing audience doesn't approach that ceiling, and it has high churn. Apple solves the reach and UX problems. Jeff provides the expertise and trust.
Monetization model: in-app creator subscriptions — the same infrastructure Apple built for Podcasts Subscriptions. Creators set their own price. Apple takes 30% in year one, 15% from year two onward. No royalty pool to manage, no rate disputes. Creators have autonomy and predictable income; Apple earns on every subscription without owning any content.
Outcomes
Results & Impact
12 weeks of real Apple Watch biometrics + workout log. The gap this feature closes.
Reflections
What I Learned
- ◆The best product ideas connect assets no one has wired together.
Apple didn't need to invent biometrics, nutrition tracking, or fitness content separately. They needed someone to argue that those three things belong in the same loop. That's a PM insight, not an engineering one.
- ◆Creator partnerships work when incentives align — not just when it's convenient.
Jeff Nippard's $50 PDF has a delivery problem, not a content problem. Apple solves the delivery. Jeff gets 100M potential subscribers instead of whoever clicks his website. The deal writes itself when both sides win.
- ◆Nutrition and recovery aren't separate features — they're inputs to performance.
The insight from the iOS 26.4 roadmap isn't that Apple is building a food tracker. It's that Apple is finally collecting the last missing variable in the adaptive training equation. This feature couldn't have been pitched convincingly a year ago.
Next Project
DentStream →