Product engineering · local-first decision system
Training that adapts when real life changes the plan.
Hypertrophy App combines fast workout logging with a deterministic decision engine: it uses the user’s own training history, constraints and recovery signals to make bounded recommendations that can be inspected, accepted and applied.
The product insight
Most training apps are good at recording what happened. The harder product problem is helping someone decide what to do next without hiding the evidence, losing the workout record or pretending the system knows more than it does.
Hypertrophy App is designed around that gap. The user keeps a genuinely useful free logger and owns their history. The deeper value is recurring intelligence: explainable training reviews, constraint-aware replanning, guarded volume adaptation and program-wide changes that remain reversible and safety-visible.
For the trainee
Less friction during the session, clearer comparisons against prior work, and a practical next step when equipment, time, soreness or missed sessions disrupt the plan.
For the product builder
A concrete example of turning domain rules into a trustworthy product loop: capture evidence, decide conservatively, explain the decision, ask for acceptance, then persist the result.
A closed-loop system users can trust
- 01 / CAPTUREMake the right action quickOne-tap set completion, previous-performance context, rest recovery, warmups, plate calculation, supersets and advanced set types support the live workout.
- 02 / INTERPRETCompare actual training evidenceWeights, reps, RPE/RIR, exercise order, muscle stimulus, breaks and history become structured signals rather than a vague “AI coach” prompt.
- 03 / RECOMMENDOffer a bounded change with reasonsProgress, hold, deload, substitute and replan actions carry reason codes, confidence and an explicit scope.
- 04 / APPLYKeep the user in controlAccepting a recommendation is separate from applying it; stale targets, rejected changes and duplicate applications are guarded.
- 05 / REMEMBERPersist the decision, not just the screenRecommendation lifecycle state travels through backup, restore, import, export and recovery paths.
Where the engineering shows
Decision logic
Goal templates, exercise equivalence, order-fatigue factors, autoregulation, deload policy and break handling create a rules-based foundation that can be tested and explained.
Product mechanics
Workout persistence, interruption recovery, setup drafts, history correction, import/export and user-owned data treat continuity as a product feature.
Commercial thinking
The free/Pro model is built around value timing: the first useful review and adaptive sample are free; recurring intelligence is the premium layer; safety warnings and owned data never disappear behind a paywall.
Validation discipline
The repository records broad automated coverage, deterministic recommendation tests and a release path. Native Apple, physical-device HealthKit and watchOS evidence are kept as explicit separate gates.
Why it is commercially interesting
This is the kind of product problem I enjoy: a user-facing experience where the visible simplicity depends on careful domain modelling, state management, decision rights, explanation and release governance underneath.
- Useful without an account, cloud dependency or generative AI claim.
- Designed for a strong first value moment: a complete evidence-backed Training Review.
- Built to turn internal algorithms into user-visible, reversible product actions.
- Clear separation between validated Expo behavior and native capabilities still awaiting device evidence.
Current status: a serious product prototype. It is not presented as an App Store launch, clinical product, production Apple Watch experience or measured market outcome.