I Built an App With AI — Now I Need Testers

00:00 — Opening

Hey, welcome back to Slow Builds.

This video is me trying to bring the channel back a bit to what I want it to be, which is a little bit about everything.

I’ve been drifting into a lot of AI, which I’m still going to keep doing, but I want to space it out so it doesn’t feel repetitive.

So this one is about a project I’m working on.

AI is still part of it because I’m using AI to help me in every part of my life, but this is about something I’m building.

It’s called Glovebox GPS. It’s not finished, but it is test-ready, so let’s call it alpha.

gloveboxgps.com is up and taking signups for testers.

01:36 — Why I Built It

This didn’t start as a business idea, a startup idea, or some next great app idea.

It started because AirTags, which I was using for tracking vehicles and other important stuff, became unreliable in certain situations.

I have two grown adult children living at home, they’re on our insurance, and they drive our vehicles.

That’s why the keys have AirTags, and the cars have them too.

It’s not about spying. It’s about wanting to know where the vehicle is, where it’s parked, and having some sense of safety if something goes wrong.

02:49 — How We Use AirTags

Usually the kids are supposed to have sharing turned on, and we also use AirTags in backpacks, luggage, and rental cars when we travel.

But one of my kids really doesn’t like turning sharing on all the time.

They’re on a long road trip right now and they do have it on because they understand the safety issue.

But when they’re home, they don’t like to keep it on.

And I get that.

They’re adults. I’m not trying to control where they are.

But if you’re taking my vehicle, I need to know where the vehicle is if something happens.

04:47 — Where AirTags Fail

AirTags don’t always work.

They depend on nearby Apple devices, good conditions, and the right environment.

If my kid is out late, around people with Android phones, or just somewhere without enough Apple devices nearby, I can wake up in the middle of the night and have no idea where the car is.

That’s a bad feeling.

What I really wanted was to know where the car was, where it went, and what happened.

AirTags tell you where something is right now. They don’t really give you the full story.

07:48 — Looking At GPS Trackers

That’s when I started thinking maybe I could build my own thing.

First I looked at regular GPS trackers.

I’ve actually liked GPS tracking ideas for a long time, even going way back to an old investment I made that tracked transport trucks.

But modern GPS trackers mostly wanted subscriptions, weird hardware installs, battery hookups, or charging cycles.

And at the end of the day they were still just trackers.

If I’m paying monthly, I want something more flexible and more useful than that.

11:15 — The Old iPhone Idea

Then it clicked.

I already have old phones.

An old iPhone is basically already a GPS tracker if it has power and data.

If that phone lives in the glove box, then I know where the car is.

And those old phones already have GPS, WiFi, LTE, batteries, cameras, maps, messaging, and internet.

I started looking at cheap SIMs and eSIMs, including ones that don’t expire, and suddenly the idea felt serious.

You plug the phone into power in the car, leave it in the glove box, and let it sleep when the car is off.

14:18 — How The App Works

The app itself is intentionally simple.

And honestly, we’re at a point now where almost anybody can build things if they can explain what they want clearly enough.

You don’t need to deeply understand code the way you used to.

The app on the phone just shows a device ID, which you claim on the website and attach to your account.

Then it tracks movement, stops, long periods of inactivity, and major location events.

It stores longitude, latitude, device ID, date, and time.

It keeps those points locally and only sends them up periodically so it doesn’t waste bandwidth or battery.

17:16 — The Problem It Solves For Me

What I like about this is that it solves a problem for me without forcing rules too hard.

I could’ve just said if you’re in the car, sharing has to be on, but I didn’t want to turn it into some big-brother thing.

This felt like a better solution.

It solves the problem I actually have, and it works within the reality of how people really live.

I still love this idea, and even though I mainly want it for myself, I do think there are other people who would get it immediately, especially if they have kids or shared vehicles.

19:24 — Why It’s More Than A Tracker

The big thing is that this isn’t just a tracker.

It’s also a backup emergency phone sitting in the car.

It can be part of the car’s emergency kit.

It has maps, internet, communication apps, translation, phone calls, and all the normal phone capabilities.

So it’s tracking in the background, but it’s also potentially useful in a real emergency.

That’s what makes it feel more valuable than a normal tracker to me.

21:02 — How AI Helped Me Build It

This is where AI helped me the most.

Not in a fake “AI builds everything for me” way, but in a very practical way.

I knew enough development to understand the structure and what had to exist, but I’d never built a Swift app from scratch before.

AI helped me figure out the right architecture, so I ended up with a React front end, a Rails API back end, and an iOS app talking into that API.

It also helped me figure out the oldest iPhone I could realistically support while still getting the app into the App Store and still collecting the background location data I needed.

25:05 — Everything Around The App

Honestly, one of the coolest parts wasn’t even the coding.

It was all the stuff around it.

I wanted to do this through my corporation, not personally, so that meant DUNS numbers, registrations, forms, approvals, emails, and Apple Developer account setup.

AI helped me through that whole process.

Then Apple denied me because I’d forgotten that I needed a proper company website, privacy policy, terms, and contact information.

As soon as I told Codex what happened, it used what I already had in Cloudflare, Railway, and my repos, built the site pieces I needed, and got everything in place.

The next day I was approved.

27:51 — Where It’s At Now

So now it’s in TestFlight, it’s in alpha, and it works.

We redid the site, put together an alpha plan, and I’ve got social accounts around it, but I’m trying to keep the whole thing pretty organic.

I do want real people using it though, because I need to find what breaks, what’s annoying, what features are missing, and what only makes sense in my own head.

AI also helped me build the signup flow, database structure, onboarding, and approval process.

There’s still lots missing like saved places, trip logging, work versus personal distance, favorite routes, and better event summaries, but the basics are there.

31:00 — Closing

So yeah, it’s still rough. It’s unfinished.

But I use it, and I love it.

And I’ve got other apps going too, which is kind of the beauty of this AI age.

You think of something, and you can make it.

Even this whole Slow Builds process now has skills, memory, and structure around it.

There’s still a motivation game to all of this though.

I almost pushed this video off until tomorrow.

But no. I’m here.

Thanks for watching.

And again, gloveboxgps.com.

Go take a look. If you sign up, let me know and I’ll get you in there.

Android’s coming.

Alright, thanks for watching. Bye.