Overview

CycleCare is an AI FemTech app I designed, built, and published solo — one of 7 production iOS apps shipped end-to-end via an AI orchestration pipeline I architected (Claude Code, Cursor, SwiftUI), with no Figma handoff and no hand-written code.

The Ask

Prototype and validate an AI-native, no-handoff design-to-code workflow using real, shipping consumer products as the testbed — not internal demos.

Constraints

• Solo: designer, developer, and publisher in one

• Zero marketing budget — growth had to be 100% organic (ASO only)

• On-device logic only for CycleCare: no runtime AI, no hallucination risk on clinical guidance

• Native SwiftUI motion and interaction quality, without a design-to-dev handoff step

What I Did — The Pipeline

Design decisions are specified directly as prompts and constraints for AI code generation (Claude Code, Cursor), which produces native SwiftUI — collapsing the traditional Figma → handoff → implementation chain into a single, designer-driven loop. CycleCare's core logic is an on-device rule-based algorithm synthesizing WHO/ACOG clinical guidelines, chosen deliberately over runtime AI to eliminate hallucination risk in health guidance.

Outcome

• CycleCare: 5.0★ App Store, ~50% page-to-install (10x industry average), 93% organic acquisition, $0 ad spend

• RelaxColors: 6.32% App Store conversion (2x category average); +52% downloads and +23% revenue from a single ASO optimisation

• 7 production iOS apps shipped end-to-end, solo, via the same pipeline

What I'd Do Differently

Standardise the prompt-to-SwiftUI-component library earlier — the first 2 apps re-solved the same interaction patterns from scratch before the pipeline matured.