ROLE:
PRODUCT DESIGNER & BUILDER
TIMELINE:
3 MONTHS
PLATFORM:
IOS
YEAR:
2025
Alarm Buddy
What if designers didn't need engineering teams anymore?
THE EXPERIMENT
I wanted to answer one question: Could I ship a complete iOS app using only design skills and AI?
Eight years ago, I had this exact idea—a gamified alarm app that made waking up delightful. I even built a design prototype. But I couldn't code, and learning Swift felt overwhelming enough that the idea stayed dormant.
With AI tools evolving rapidly, this felt like the perfect moment to test whether designers could finally bridge that gap. Not just prototype, but actually ship. The alarm app wasn't just a product test—it was proof of concept for a new way of working.
THE BUILD
Design as Code
I started the way I always do—defining the problem, sketching ideas, building a design system. But instead of handing off specs, I worked with AI to translate them directly into Swift code.
The workflow looked like this:
Design a feature in detail (alarm setup flow, game mechanics, character system)
Describe it to Cursor with context about iOS patterns
Review generated code, test on device, iterate
Debug issues, refine interactions, ship
What I Built
Core Experience: Three buddy characters (Biscuit, Luna, Pebble), each with unique themes—colors, sounds, visual identity. Users pick their buddy, set an alarm, choose a mini-game.
The Games:
Find the Buddy: Spot your character among similar options
Catch the Buddy: Tap the moving target
Both hit the sweet spot—cognitively engaging without being frustrating. Wake you up effectively, but feel like play.
Design System: I kept the app shell clean and minimal—letting the interface breathe. Then made the gamified elements vibrant, pastel, kawaii-themed. This contrast was intentional: the app feels calm and functional until the game kicks in, then it's pure delight.
What I Learned
Building forced me to understand iOS constraints in ways design specs never could:
Technical Deep Dives:
iOS's 64 notification quota limit and workarounds
SwiftUI state management and layout systems
App Store requirements and TestFlight distribution
AI Orchestration:
Used Claude to refine prompts iteratively—learning to describe problems better
Experimented with different models based on task: Sonnet 4.5 for architecture decisions, GPT-4 for specific implementations, Gemini 2.5 and Grok-code for alternative approaches
Learned when to switch models vs. when to refine prompts
Problems I Solved
iOS Notification Limits: iOS caps apps at 64 scheduled notifications—brutal for an alarm app where users might set multiple daily alarms. My solution: strategically trigger 5 main notifications at 30-second intervals (iOS's max audio duration), balancing reliability with UX. This approach reliably handles ~15 alarms while staying within quota.
Built smart notification cleanup to manage the quota—removing old notifications, prioritizing upcoming alarms.
Bypass Prevention: Early testers tried force-closing the app to skip games. Added a re-engage notification that fires if users attempt to bypass gameplay—bringing them back to complete the challenge.
Character & Animation: Used character animations from LottieFiles (simple license) and designed the gameplay mechanics around their movements and expressions. This constraint actually improved the games—I had to think creatively about how each character's animation could become core to the interaction.
THE PRODUCT
After 3 months, I shipped a focused, delightful alarm experience. Not feature-bloated, not trying to be everything. Just: set alarm → pick buddy → play game → start your day with a smile.
What's Live:
3 buddy characters with unique themes
2 mini-games optimized for morning engagement
Clean, aesthetic interface
Reliable alarm system that respects user preferences
What's Next:
Monetization (premium buddies, new games)
Sleep wellness tracking
Custom Sound Selection
I intentionally shipped lean. The MVP proves the concept works. Now I can layer in complexity based on real user feedback instead of assumptions.
WHAT THIS MEANS
This project isn't just about shipping an app. It's about what becomes possible when designers control the full stack.
Speed of Validation: I compressed 6 months of traditional product development into 3 months, solo. No sprint planning, no handoff delays, no "engineering says it's not possible."
Learning Compounds: When you can test ideas immediately, learning accelerates. I learned more about iOS architecture in 3 months of building than years of handing off specs.
Creative Control: Every detail—from game timing to character animations to notification sounds—matches my original vision. No compromises forced by technical debt or resource constraints.
New Designer Capabilities: I'm not suggesting designers should replace engineers. But we no longer need to wait for engineering resources to validate ideas. We can prototype, ship MVPs, gather real user data, then bring in engineering partners to scale.
The future isn't "designers who code." It's designers who orchestrate AI to bridge the gap between concept and reality.
TRY IT YOURSELF
Alarm Buddy is live on the iOS App Store. Download it, wake up to something delightful, and see what a designer + AI can build in 3 months.
If this approach resonates with you, I'm happy to talk about it.
