Design AI Stack

Design AI Stack

Build

Don't sleep on Cursor's Plan Mode

Refactoring an app as a designer (without knowing engineering)

Sherizan's avatar
Sherizan
Jan 16, 2026
∙ Paid

I vibecoded an app. It worked. Then it didn’t.

About a year ago, I launched Nasma, a daily habit app.
Not after months of planning. Not with a big team.

I tried Cursor for the first time in Jan 2025, wanted to see if I could actually build a native iOS app, and… two weeks later it was live.

I posted it on LinkedIn.
People downloaded it.
Feature requests rolled in.

So I did what every vibecoder does:

  • ship fast

  • add features

  • keep momentum

Today:

  • thousands of active users

  • users mainly from the US, UK, India, Indonesia

Except retention was bad.

After Day 14, only ~5% of users came back.

That’s when I stopped shipping and started thinking.


The mistake I made (classic vibecoder move)

I assumed:

“If something feels slow or messy, I just need to refactor everything.”

Wrong.

What I actually needed was clarity before code.

That’s where Cursor’s Plan Mode quietly became the most useful feature I wasn’t using properly. (Well it also didn’t exist back in Jan)


How I actually use Plan Mode (not the marketing version)

I don’t use Plan Mode to “build”.

I use it to:

User's avatar

Continue reading this post for free, courtesy of Sherizan.

Or purchase a paid subscription.
© 2026 Sherizan · Privacy ∙ Terms ∙ Collection notice
Start your SubstackGet the app
Substack is the home for great culture