Detailed Case Study
Drop It: The AI-Agent & Biometrics iOS App
A menstrual cycle and workout synchronization system that dynamically adapts training plans based on daily physiological biofeedback.
The Context & Product Vision
Drop It was designed to bridge the gap between female biological cycles and strength training, helping athletes optimize their routines according to cyclical shifts in performance, symptoms, and biofeedback. The vision required a system that felt alive, adapting workouts dynamically in response to real-time body metrics.
The Core Systems & Engineering Solutions
Menstrual Cycle & Biofeedback Engine
Designed the core algorithmic pipeline that maps hormone cycle phases and processes physical symptoms. The system runs calculations on daily user inputs to project physiological fatigue limits, automatically adjusting training intensity and volume recommendations.
Adaptive Workout System
Architected a robust workout coordinator that handles exercise classifications, manages set and repetition schemes, and coordinates real-time adaptations of active training sessions based on physiological metrics.
Agentic AI Orchestrator
Engineered a custom class utilizing Swift asynchronous streams to manage real-time, conversational interactions with LLMs. The orchestrator decodes streaming JSON payloads on-the-fly, dynamically translating agent decisions into UI actions: rendering token-by-token conversational text, mutating workout card states (modifying exercises, volume, or loads), or triggering automated generation flows.
Apple HealthKit Integration
Built a comprehensive, asynchronous manager to handle OS-level permissions and query physiological parameters. It parses active energy burn, heart rate variability, and sport logs, translating them into structured variables within the app, while writing completed training sessions back to the health database.
SEO & Growth Infrastructure
Designed a multi-channel growth funnel utilizing Next.js, Vercel Web Analytics, Google Search Console, and Meta Ads event tracking:
- Funnel Landings (Bottom of Funnel): Custom static landing pages injecting structured JSON-LD schemas to rank on transactional searches.
- SEO AGI Optimization (Top of Funnel): Designed MDX articles and guides optimized for Generative AI search crawlers (ChatGPT Search, Perplexity), tracking crawler user-agents directly in server logs to measure LLM referencing.
- Performance: Achieved a CPI of ~0.80€ and a CAC of ~13€, driving over 5,000 installations.