Let’s be real for a second — I love building things. Side projects are fun in theory. But in practice? They were slowly draining the life out of me.
You finish work, open your IDE with big dreams, and next thing you know, you're three hours deep into setting up a reverse proxy, writing the 15th migration script of your life, or fighting with some obscure Docker error that makes you question your career choices.
I wasn’t burned out by coding.
I was burned out by everything around it.
Then Codex (and AI coding tools in general) came along — and suddenly, side projects didn’t feel like unpaid overtime anymore.
The Burnout of “Build It All from Scratch”
Every dev hits this point eventually:
You have an idea. It’s a cool one. You want to get started.
But before writing even a single line of logic, you’re already buried under:
-
Choosing the stack (again)
-
Setting up the repo and CI/CD
-
Configuring Docker and environment variables
-
Creating a database, seeding it, writing auth from scratch, etc.
The overhead kills the vibe.
Especially when you’ve done it a dozen times already — at work, in past projects, or for tutorials that never went beyond Hello World.
The irony? The thing that should be exciting — building your own thing — ends up feeling like a chore. So you drop the project… again.
AI-Assisted Dev Flow — How I Build with Codex
Codex doesn’t build the perfect solution from scratch. It’s not magic.
But here’s what it does do:
-
Generates working scaffolding fast
-
Writes solid first drafts of services, endpoints, migrations, unit tests
-
Explains syntax and helps debug without having to tab out
-
Remembers your context (usually) so you’re not repeating yourself
Now I can throw ideas at it, get a decent starting point, then just tweak or improve. I’m still in charge — but I’m not starting from zero every single time.
The 70/30 Rule: AI Drafts, I Refine
These days, 70% of the boring or boilerplate code in my side projects comes from Codex.
But 100% of it is reviewed, cleaned up, and tested by me.
And that’s the sweet spot.
AI gives me speed, but I keep the quality. I don’t worry about perfection in the first pass anymore — I iterate. The whole thing feels lighter, faster, and more fun.
Side note: it's weirdly satisfying to refactor AI-generated code. Feels like teaching your intern… who just happens to write C# at 3am without complaining.
Case in Point: Kvblog Search
When I was writing a blog post for Kvblog, I casually asked Codex to add a search endpoint. While I focused on writing, it drafted a usable API method with filtering logic, query parameters, and a DTO. I cleaned it up, added pagination, wired it to the frontend, and just like that — search was done.
No context switching.
No breaking flow.
Just quick prompts and progress.
Wrapping Up
I still love coding.
But I no longer feel the need to prove myself by suffering through every setup step alone.
Codex doesn’t just write code — it brought the joy back into building.
Side projects feel like play again, not work in disguise.
So yeah — side projects were killing me.
Now I build faster, ship more, stress less.
AI and chill. Let’s keep it going.
Disclaimer: This post was generated through AI based on the content of my discussions with
Chatgpt to add some initial content to the blog. A more sophisticated “Lorem ipsum” in the era of AI.