I Stepped Away From Coding for Years. How AI Helped Bring Me Back?
I’ve been building iOS and Android apps for over 15 years. ExtraMile — a mileage tracker for gig workers and self-employed drivers — had been live for over a decade. Then life got in the way, I stepped back and couldn’t maintain it. The app stayed up, but development stopped.
When I finally came back to it, I expected to pick up where I’d left off. Instead, I found a programming world had totally changed.
Coming Back to a Changed World
The codebase was still mine. The location engine I’d spent years perfecting still worked. The app still tracked drives, still calculated IRS deductions, still did everything it was supposed to do.
But the tools around me had completely changed. AI coding agents were everywhere. Every developer forum I visited was talking about them. I was skeptical — this is my code, my app, years of careful work — and honestly a little defensive about it.
I tried it anyway.
The Rules I Set for Myself
The first thing I did was draw a hard line: AI would never touch my location engine.
That engine is the heart of ExtraMile. It’s the part that decides when you’re driving versus walking. It handles the edge cases — the traffic stops, the parking lot crawls, the trips that start and end in weird ways. I’d spent years getting it right. I wasn’t about to let an AI make aggressive changes to something that just works.
Everything else, though, was fair game.
The model I settled on: AI does the heavy lifting, I do the guiding. When it drifts from what I want, I pull it back. I stay in the loop on every decision. I’m not outsourcing my app — I’m working with a very fast, very capable collaborator that needs direction.
That distinction matters. The best results came when I had a clear picture of what I wanted and used AI to execute it quickly, not when I asked it to figure out what I should build.
What Actually Changed
ExtraMile already had good bones. The core features were solid. What it needed was a modern UI — and that’s where everything shifted.
PDF Reports: A Week of Work in Minutes
The old report output was functional but rough. I wanted polished, well-formatted PDF reports that users would actually want to send to their accountants.
Designing and implementing that from scratch — getting the layout right, the typography, the spacing, making it look professional — would have taken me a week of back-and-forth. With Claude Code handling the implementation while I directed the design, it took minutes to get to a first version I was happy with.
CalendarView: A New Feature in an Hour
I wanted to add a calendar view so users could see their trips mapped out by day, make sense of their mileage at a glance. New view, new design, SwiftUI implementation.
Start to shipped: about an hour.
The part that surprised me wasn’t the speed — it was the quality. I didn’t have to compromise. I got the layout I had in my head, with the right visual hierarchy, implemented correctly.
Dashboard UI: 100 Iterations, No Designer
This is the one that still amazes me.
The dashboard is the first thing users see. Getting it right matters. Before, iterating on a UI like that meant either hiring a designer or burning weeks of my own time going back and forth. Both options have real costs.
With AI in the loop, I could iterate hundreds of times — try a layout, see it, say “the trip cards need more breathing room” or “that cost figure needs to be bigger,” and see the change in minutes. I could explore directions I would have never had time to try before. The final dashboard is better than anything I would have shipped working alone, and it got there faster.
Drive Score: Debugging Gets Faster Too
Drive Score — the feature that gives users a rating on their driving behavior after each trip — had a bug I’d been circling. Having AI in the debugging loop, talking through the logic, testing hypotheses quickly, cut the time to resolution significantly.
What the Right Balance Looks Like
I want to be honest about what this workflow actually is, because I’ve seen a lot of hype and a lot of backlash, and the truth is somewhere more nuanced.
AI did not build ExtraMile. I did.
The location engine — the thing that makes the app actually work — is still 100% mine. The product decisions, the feature priorities, the design direction, the judgment calls about what users need: all mine. The 15 years of understanding iOS location APIs, CoreData, how apps behave in the background — that didn’t go anywhere.
What changed is the cost of execution on the UI layer. Things that used to take days take hours. Things that used to take hours take minutes. I can try more ideas, ship faster, and spend less time on implementation and more time on judgment.
That’s the right framing: AI as a force multiplier for an experienced developer, not a replacement for one.
The App That Came Out the Other Side
ExtraMile launched on Product Hunt with a completely rebuilt interface: a new dashboard with Map, Behavior, and Summary tabs; a calendar view; polished PDF reports; real-time drive scores; and an onboarding flow that actually explains the value of each permission before asking for it.
The core is the same app it always was — the location engine that tracks every deductible mile, free forever, with no data lockout. Pro adds automatic monthly reports delivered to your email, one-time purchase, no subscription.
It’s the best version of the app I’ve ever shipped. And I genuinely don’t think I could have gotten here without being willing to change how I work.
For Other Indie Devs
If you’re skeptical — good. Skepticism is the right starting point. Know what you’re not willing to hand over. Know where your expertise lives and protect it. Use AI where it can move fast without risk, not where mistakes are expensive.
The developers who are getting the most out of these tools aren’t the ones using AI to build everything from scratch. They’re experienced people who know exactly what they want and are using AI to get there faster.
That’s a different skill than traditional solo development. It takes practice. But once it clicks, it changes what’s possible for a one-person team.
ExtraMile2 is available free on the App Store. Automatic mileage tracking, drive scores, unlimited trip history, IRS deduction calculations. Pro is a one-time purchase for automatic monthly reports.
Questions or feedback? Contact us