The Spark: A Personal Mandate
My husband and I discovered that shared challenges strengthen our connection in meaningful ways. Yet tracking these commitments across scattered notebooks and fragmented messages proved frustratingly inefficient.
This everyday friction sparked the creation of Love2 — a streamlined, centralized, and private application designed specifically for couples navigating shared goals together.
The Experiment: "Vibe Coding"
I embarked on testing a novel workflow I call "Vibe Coding" — where my role fundamentally shifted from designer to director. Rather than crafting detailed mockups manually, I authored a comprehensive technical and design specification to serve as the AI's creative blueprint.
This specification became the "source code" for the AI developer (Lovable), tasking it with executing my complete vision from concept through implementation — an end-to-end handoff that challenged traditional design processes.
The Process: Execution, Hurdle, & Pivot
Initial Success
The AI flawlessly executed my front-end vision, rendering the application with the precise coral-to-pink gradient and structural architecture I specified in the brief.
The Technical Hurdle
Complex cloud-based features — multi-user authentication and the "dare" tracking system — exposed a critical limitation when the AI's cloud deployment infrastructure failed repeatedly.
The Director's Call
I made a strategic pivot, directing the AI to build a client-side-only version. This allowed validation of the complete user experience without backend dependencies.
The Final Prototype
Client-Side PWA ImplementationFollowing the strategic pivot, the AI successfully generated the client-side Progressive Web App. The final prototype validates the user flow exactly as originally envisioned — from challenge creation to photo sharing and progress tracking.
The application demonstrates how thoughtful direction can guide AI tools to produce functional, beautiful experiences even when working within technical constraints.
Reflection & Conclusion
The experience proved that AI can be a remarkably powerful developer for executing clear design specifications. However, it simultaneously exposed significant limitations in handling complex backend infrastructure and cloud services.
This solidified a crucial insight: the designer's role is evolving into that of a director — providing essential human-centered vision, guiding technical execution, and strategically navigating inevitable trade-offs to deliver viable products.
What AI Did Well
- Executed front-end vision precisely
- Rendered complex UI from text specs
- Adapted quickly to the pivot
- Generated working PWA code
Where AI Fell Short
- Cloud deployment infrastructure
- Multi-user authentication
- Complex backend systems
- Debugging its own failures