Stepping Away from the Terminal

I haven't touched my laptop in a few weeks.

That sentence sounds wrong, even to me. I'm a software developer and I've recently been building for Apple platforms, which means building on a Mac. Not optional. The entire development workflow is tethered to macOS.

Except lately, I haven't needed to sit in front of one.

There's a script running on my Mac right now:

while true; do
    claude remote-control;
done

That's it. The machine sits in the basement, screen off. I start Claude Code sessions from my phone, and the Mac picks them up. Builds install on my phone over home Wi-Fi. I review diffs on my phone, describe what I want, and watch the commits land.

This didn't happen overnight. It happened in stages. Each one feeling like a small step and looking like a large one in retrospect.

Four Steps Back

In September, I switched from GitHub Copilot to Claude in VS Code. That was the obvious move. Better completions, better understanding of intent, better at holding context across files.

In November, I closed VS Code entirely. Claude Code in the terminal felt more natural than any editor integration. Simpler interface, fewer distractions. I still had the code on my machine, but I stopped looking at it in an editor. The code started feeling like a side effect of the conversation, not the main event.

In January, I moved to Claude Code on the web. The conversation wasn't even on my local machine anymore. But I still needed Xcode open to build, run, and validate. I completed multiple tasks throughout the day, but at the end of it I still had to verify the build on the laptop.

In February, that last thread snapped. claude remote-control let me drive the Mac from my phone. The app built and installed over Wi-Fi, and I verified it right there. No IDE open. No terminal open. No laptop screen on.

Each step felt like giving up control. Each step turned out to be giving up friction. A phone and an idea became enough to start building.

What I Was Building

The app is called Foveate. My wife showed me an Instagram reel of someone speed reading, and I'd never heard of RSVP before. It intrigued me, so I started exploring. Research, specs, then code. Built in two days, then used it for two weeks as my daily reader.

Then I noticed the app didn't fit my actual reading habits. I didn't need a speed reading app. I needed a way to keep up with the engineering world, where speed reading was just one tool in the box.

The product needed to change. And I didn't need a laptop to change it.

Building from the Phone

I did the pivot entirely from my phone.

I was on the couch, thinking about the app, and I just started. Opened Claude Code on my phone, explained the new vision, and Claude captured it in a spec and started implementing.

The workflow felt ordinary. I described what I wanted, Claude Code made the changes, built the app, and launched it on my phone. I tested the experience, gave feedback, and it fixed things and relaunched. That was the dev loop.

Specs Made This Possible

I wrote about spec-driven development a few weeks ago. Foveate is where that workflow proved itself.

Developing from a phone means no browsing code. No visually scanning a file to understand its structure. No holding a mental model of the codebase by scrolling through directories.

But I can read a spec. And when the spec is clear enough, the code becomes a detail I don't need to hold in my head.

The initial vision of Foveate was expansive. 18 feature specs covering PDFs, ebooks, importing from open libraries, and more.

When I realized building everything around RSVP wasn't what I wanted, the specs made the pivot tractable. I described the new direction to Claude: an RSS reader with articles I could skim through and LLM-generated digests to help me catch up. Claude looked at the spec index, figured out which specs to cut and which to add, and planned the path from the old vision to the new one.

Without specs, a pivot like that would have meant re-reading code to understand what existed. With specs, it was a conversation about intent.

What This Feels Like

It feels like dictating a letter to someone who types faster than me. I'm still the one deciding what to say, still the one who knows the recipient and the purpose. The mechanical part has been absorbed. My job shrank to the part that was always mine. The thinking.

I don't want to oversell this. Reading diffs on a phone is not pleasant. Complex architectural decisions still benefit from a larger screen. Debugging a subtle layout issue is easier when I can inspect the view hierarchy directly. There are real limitations.

But the default has shifted. The phone is enough most of the time, which means I can build whenever inspiration strikes. The laptop has become the exception, not the primary tool.

The Real Story

None of the individual parts are remarkable. AI-assisted coding exists. Remote development exists. Product pivots happen all the time.

What's notable is how they combine to make software development more accessible. Each step away from the traditional workflow removed a dependency I'd assumed was essential. No specific machine, no specific editor, no desk required. A phone and an idea turned out to be enough.

The barrier to creating software is lower than it's ever been. I'm curious what I'll build next from the couch.