When I Stopped Building Everything

My first prompt for Beads of Hope T1D was not exactly modest.

I asked the AI to build a “fully functional, production-ready ecommerce and community website” for a philanthropic organization selling handmade bead products to raise money and awareness for Type 1 Diabetes.

In plain English: I asked for the whole thing.

A shop. A customize page. A community area. Checkout. Products. Branding. The works.

About 20 minutes later, I was staring at something that looked like a real storefront. It had a shop page. It had Google login. It had Stripe checkout connected in some way. It had more database tables than I knew what to do with. Twenty-two of them, apparently.

I remember feeling thrilled and completely disoriented at the same time. Like I had asked someone to sketch a lemonade stand and they dropped off a shopping mall.

At first, that felt exciting. Then the first wall showed up.

That same evening, none of my uploaded product images were visible on the published site. There were duplicate category buttons on the shop page. Every time I clicked where a product image should have been, the screen snapped back to the top of the “all” category.

I kept telling the AI some version of, “The images are still not visible, and every time I click the box where the image should be, it jumps back to the top.”

And then I told it again.

And again.

For hours.

That was the first time the direction changed for me. Not in a dramatic “delete half the app” way, but in a quieter, more important way: I stopped trying to build everything at once.

I had started with this giant vision of an ecommerce and community website. But the actual work was not giant and glamorous. It was one missing image. One broken click. One duplicate button. One product page that behaved strangely.

A few days later, the customize page threw an error that said: Cannot find module '@/assets/blue-ribbon.png.asset.json'.

I did not know what an asset.json was. I still barely do. I just knew the page was unhappy, and I was unhappy looking at it.

That was another little pivot. I had to stop pretending I understood the whole machine. I started treating the build more like a conversation with a very fast, very literal assistant. I didn’t need to know everything. I needed to notice what was wrong, describe it clearly, and decide what mattered next.

And what mattered next kept getting smaller.

Move the blue ribbon down a quarter inch.

Make the T1D Hope text stand out more with white shading.

Try a hover-zoom on the badge reel photos. No, 3X is too much. Try 2X. No, only on the badge reels.

These sound tiny, and maybe they are. But at the time they felt huge. I wanted the site to feel like a real brand, not something I had duct-taped together during a late-night burst of optimism.

Then the security scans came back, and that changed the direction again.

Suddenly I was reading about guest cart hijacking and client-controlled prices. One issue meant someone could write to a guest cart when they shouldn’t. Another meant a customer could technically send their own price if the order function trusted the wrong thing. There were also RLS policies, which are database rules that decide who can read or change what.

The AI fixed them, but I would be lying if I said I felt confident reading that scan output. I felt out of my depth. Very out of my depth.

That pushed me into a more serious version of the project. This wasn’t just “make it pretty” anymore. If people were going to place real orders with real money, I had to slow down and take the boring scary parts seriously.

By June 9, the AI told me the code-side work was essentially complete. That should have felt like a finish line.

Instead, it felt like a handoff.

I still had to set up Stripe with real bank information. Real shipping. Real tax. Real account activation. That was when I realized the app being built was not the same thing as the business being open.

The builder could help me make the store. It could not be brave for me.

So the pivot was this: I stopped chasing the biggest version of the idea and started working toward the simplest version that could honestly go live.

A real shop. Real products. Real checkout. Real purpose.

The community dream can grow later. The polish can keep improving. The launch video can be adjusted from “She was four when everything changed” to “He was two when everything changed,” because that is the real reason I’m doing this.

It’s for my son.

And that made the decision easier. Not easy. Easier.

I didn’t need to launch every idea I had. I needed to launch the part that could begin.

Small takeaway: Sometimes the pivot isn’t changing the dream. It’s making the dream small enough to actually carry.