I don’t treat AI like some magic tool that writes all my code. It’s more like a fast assistant that helps me move through development without getting stuck on small things.
Over time, I built a simple workflow where I use AI at different stages of building apps — mostly Flutter + Supabase projects — to save time and stay focused.
Table of Contents
- 1. Thinking Before Coding
- 2. Faster Boilerplate
- 3. Debugging Without Losing Time
- 4. UI Decisions
- 5. Thinking About the Product
- How it all comes together
- Final thoughts
1. Thinking Before Coding
I usually start by explaining what I want to build in plain English.
Then I use AI to help me:
- break it into smaller features
- figure out what I might be missing
- organize the structure before I touch code
This step helps me avoid messy architecture later.
2. Faster Boilerplate
Let’s be honest — a lot of coding is repetitive.
Things like:
- basic screens
- models
- API setup
- state management scaffolding
I let AI generate a starting point, then I adjust everything to match my project.
I don’t copy-paste blindly — I treat it like a rough draft.
3. Debugging Without Losing Time
When something breaks, I don’t immediately start guessing.
I just:
- paste the error
- explain what I expected
- ask what could be wrong
Most of the time, AI points me in the right direction faster than searching randomly.
It doesn’t always give the exact fix, but it narrows things down a lot.
4. UI Decisions (Because I’m Not a Designer)
I’m not a UI/UX designer, so I use AI to sanity-check my ideas.
I ask things like:
- is this layout readable?
- what could improve spacing?
- does this color combo make sense?
Then I implement it in Flutter using my own design system.
It’s basically feedback, not auto-design.
5. Thinking About the Product
Sometimes I get stuck only thinking as a developer, not a product builder.
So I use AI to ask:
- what features are missing?
- what would make this app more useful?
- what would users expect here?
This helps me think beyond just “does it work” and more about “does it feel complete.”
How it all comes together
- Plan with AI
- Break into features
- Build base structure
- Refine manually
- Debug with AI when needed
- Improve UI/UX
- Repeat
Nothing complicated.
Final thoughts
AI doesn’t replace development skills — it removes friction.
If you understand your code, it becomes a powerful tool. If you don’t, it becomes a shortcut that slows you down later.
For me, it’s somewhere in between — a helper, not a replacement.
I’m curious how others are using AI in their workflows — especially in real projects. Is it more for debugging, planning, or just boilerplate generation?
Top comments (1)
I actually forgot to mention—I once spent 2 hours debugging a Supabase policy before the AI pointed out a typo in 2 seconds. Has that happened to anyone else?