The Spark: How a Curious Experiment with AI Became a Real Travel Product
Building Sunday Planning: A Founder’s Journey Into AI-Assisted Product Development (Part 1)
2025 didn’t start the way I expected. AI was everywhere, but most of it felt like entertainment—funny screenshots, novelty demos, friends showing me clever prompts. I was skeptical. It all felt like a toy.
But then it started creeping into my everyday life.
ChatGPT became a faster way to Google things when I didn’t want to sift through search results. I used it to understand tiling patterns and Schluter systems during my home renovation. I generated themed party invitations. I even experimented with landscaping mockups to imagine what my backyard could look like.
And then, unexpectedly, I was laid off. I used AI to rewrite a 20-year-old résumé and produce cover letters tailored to different roles. The results were shockingly usable.
But I still wouldn’t have said AI was “life changing.”
Not yet.
The Moment Curiosity Took Over
I’ve spent my career in product management, so naturally I started wondering:
Could AI help me build a real product? Not just documents—software. Something that actually solves a problem?
That question led to the earliest concepts that would eventually become Sunday Planning, a tool to help travellers plan trips, save places, and share recommendations with friends.
I began testing the new wave of “AI app builders” like Lovable and Base44—platforms that promised to generate full apps from a paragraph.
They were magical… at first.
But the moment I wanted to change anything, the whole thing unraveled. Every small adjustment felt like pulling on a loose thread: the whole sweater came apart.
Cursor came next—an AI embedded into a development environment. The idea was brilliant, but I didn’t yet have the technical foundation to debug the code it generated. I asked too much of it, and it asked too much of me.
So I made a decision: If I wanted to build Sunday Planning, I had to actually learn modern web development.
Using AI to Learn, Not Replace Learning
ChatGPT and GitHub Co-Pilot became my tutors. They explained patterns, showed examples, clarified documentation, and helped me build an intuitive sense of how everything fit together. I watched countless videos, asked countless questions, and slowly shrank the overwhelming breadth of the modern stack.
Eventually, I committed to a lean, powerful setup:
Next.js + Supabase.
Fast, clear, capable of scaling, and beginner-friendly.
It still felt impossible to build a full application alone—but no longer unimaginable.
And then Claude Code arrived.
The Door Opens
Claude Code didn’t feel like a chatbot. It felt like a capable engineering partner who could reason, plan, and explain. It wasn’t magic—but it was the first time AI felt viable as a collaborator on a real software product.
I didn’t yet know it, but this was the turning point—not just for Sunday Planning, but for how I thought about building things entirely.
And the real learning was still ahead.
A Message for Travellers, Builders, and the Curious
This first phase of the journey wasn’t about writing code.
It was about rediscovering curiosity and realizing that AI isn’t just a shortcut—it’s a catalyst.
In Part 2, I share how I learned to “teach” AI the product I wanted to build—and how that changed everything.
