The App That Changed My Life in 1 Month: Remodex — The Codex for iOS
It started because I couldn't sit still without my Mac.
Every time I left the desk, I felt this itch. There was always one more thing to ship, one more bug to fix, one more idea I wanted to prototype. But I was stuck waiting until I got home.
I needed a way to keep coding from my phone.
Then Claude rolled out /remote-control.
That was exactly what I needed. I could keep building from anywhere, straight from my phone, talking to my agent like I was at my desk.
So I went on X to see who was doing the same thing for Codex.
That's when I found a post from @SIGKITTEN.
He had built the primordial Codex iOS app in a few hours using the Codex App Server. Rough, early, but it worked.
That was the spark.
The first version wasn't even called Remodex. It was Phodex.
Phodex was supposed to be a 24/7-without-mac app.
VPS + Traefik + Docker + GitHub. A whole environment running in the cloud where you could code and send PRs straight from your phone, no Mac required.
I was deep into building it.
Then 2 weeks in, @ajambrosino from OpenAI dropped a post leaking Remote Connections.
I read it twice.
It was clear what was coming: OpenAI was about to release something native and official. Their own iOS app for Codex. Just a matter of time.
And honestly? That's the moment something flipped in my head.
That night I made a decision. And it had nothing to do with strategy.
It was March 8th.
I didn't pivot because I was scared of getting steamrolled. I didn't pivot to "save the project."
I pivoted because I thought it would be really, really funny to ship the Codex iOS app before OpenAI did.
That was it.
Not a business decision. Not a competitive move. A personal challenge, half meme, half for the story I'd get to tell one day.
The kind of thing you do just so that years from now you can casually drop, "yeah, I shipped the Codex iOS app before OpenAI did" and watch people's reaction. Maybe tell my kids one day. Maybe just laugh about it with friends.
So I scrapped Phodex. The cloud setup, the Docker pipelines, the GitHub integration. All of it.
I decided to build the Codex iOS app instead.
Just me, my Mac, and a deadline I set in my own head for the meme of it: finish it in one night.
I posted about it on X around midnight. Then I started brainstorming and building with Codex.
Around 4am, it was done.
The first working version of Remodex.
I closed the laptop, opened a stream of the F1 Grand Prix, and tried to come down from the build high. A few hours later @romainhuet followed up with kind words.
That's when I realized this was actually going somewhere.
Working with the Codex App Server was a mix of "I love this" and "why is this so painful."
What I genuinely love about OpenAI is that they ship these tools and let us build on top with our own subscriptions. Open source primitives that actually work.
But it wasn't so easy at first.
- Getting the tool calls right
- Keeping responses in the correct order
- Parsing everything cleanly
- Live updates that didn't break under load
Every one of those was its own battle.
Credit where it's due though: they recently listened to the feedback and made the Codex App Server a bit more responsive.
But I figured it out. And what came out of it was something I was actually proud to ship.
Here's the part nobody likes to hear.
I took this challenge knowing exactly how it would end.
It was always just a matter of time before OpenAI released their own version. I wasn't naive about it. I just decided the journey was worth more than the destination.
Today, exactly 1 month after the first AppStore release, @ajambrosino leaked that they're working on the official Codex mobile app.
There it is.
The success of Remodex is going to come to an end. I knew it. I built anyway.
But I try to look at it differently.
Because here's what I actually got out of it:
- The eyes of many people inside OpenAI
- Real conversations with their team
- A compliment from @theo, including on the fact that it's open source
- Roughly 4k new followers in 1-2 months
Those are not small wins. Those are the kind of wins you can't buy and can't fake.
You build something good, in public, and the right people notice.
Now I'm working on something new.
It's called dpcode.cc, a clone of T3Code with way more features and a better UI.
And I've been thinking: what if I connect it back to Remodex?
Expand Remodex beyond just Codex. Add Claude. Add OpenCode. Add Cursor.
Make it the remote control for any agent you want to use, from anywhere, on your phone.
That's still up in the air. But the idea is sitting there, getting louder every day.
The lesson I'm taking from all of this:
Don't wait for the perfect window to ship something. Don't refuse to build because a bigger company might do it later. They probably will.
But if you build it first, in public, and you do it well, the doors that open along the way are worth more than the product itself.
Remodex was never just about the app.
It was about putting myself in rooms I had no business being in. And it worked.
I'm documenting the full journey on X/Twitter, including what I ship, what breaks, and what I learn along the way.