You Built an App This Weekend. Now What?

Dramatization. Sort of.

Our CTO, Jamon Holmgren, gets DMs. A lot of them. Someone reaches out. They built an entire app over the weekend. Auth, dashboard, real-time updates, the works. It looks like a real product. It is a real product. Jamon is impressed. He says something nice.

There’s a long pause. The typing indicator appears. Disappears. Appears again.

“So… I was kinda hoping you could actually take a look at it? Like, just real quick?”

If you’ve ever been the friend who’s “good with cars,” you know exactly how this conversation goes. If you’re a senior developer, you definitely know. For years, everyone in your life came out of the woodwork with their crazy app ideas, begging you to build them. Today? They don’t ask you to build it anymore. They already built it. Now they ask you to fix it.

It happens to Jamon more than you’d think. The tools change (Replit, Bolt, v0, Lovable, Cursor, Claude) but the question never does.

Can you help me get this thing across the finish line?

“Claude, please fix this”

Stages of grief.

Some of these people actually do get across the finish line. They shipped. It’s in the app store. Customers love it. Downloads are climbing.

But now they’re trapped.

Every bug, every feature request, every tiny change is another conversation with Claude. “Claude, please fix this crash.” “Claude, why did the login break?” “Claude, a customer is begging for dark mode and I don’t know where to start.” Claude accidentally deleted something important and they don’t know how to get it back. Or even what’s missing.

They didn’t ship a product. They shipped a conversation with an AI that they can never complete.

The AI will store session tokens in plaintext. It’ll say “LGTM” when the screenshot is missing a button. It doesn’t know what it doesn’t know.

The gap

The back half took a weekend. The front half takes the rest of the year.

AI gets you 80% of an app. The last 20% is where the actual engineering lives — and you can’t prompt your way through it.

That last 20% looks like this:

• App store compliance. Apple will reject your app for reasons you didn’t know existed. Privacy manifests, entitlements, review guidelines that change quarterly. AI doesn’t know what Apple rejected last week.
• Performance. 60fps scrolling, startup time under 2 seconds, memory leaks, battery drain. It works fine on your phone. It doesn’t work fine on your users’ phones.
• Security. Auth flows, token storage, certificate pinning. The difference between “works” and “safe” is invisible until it isn’t.
• Architecture. No tests, no deployment pipeline, no accessibility, no clear separation of concerns. The code works today. Tomorrow it’s a haunted house.

At Infinite Red, we’ve spent over a decade helping teams ship real mobile apps.

We’re not guessing about this.

A client came to us a while back — big app, millions of users, real revenue. Their team consisted of native iOS and Android developers who were simultaneously learning React Native and agentic coding tools.. They leaned hard into AI acceleration. And it worked. Sort of.

We put five engineers on AI code review for months. Even Claude plus a competent developer would look at this code and say “yeah, that’s fine.” It wasn’t fine.

Eventually, with our help, the team hit a groove. Training, documentation, linting, CI/CD guardrails, foundational architecture — all the boring stuff nobody wants to build. Now they’re shipping to the app store on their own. But the AI didn’t build that runway. We built it together.

You won’t see this part in the demo. Nobody’s tweeting “month five of reviewing AI-generated pull requests and we’re still finding bugs.” The messy middle doesn’t make the highlight reel. But it’s where the actual product gets made.

We’ve literally always done this

Hey, this looks familiar!

Someone has to do this work. That’s been our whole job for over a decade.

For years, our version of this was Ignite — a battle-tested React Native boilerplate, 19.7k GitHub stars, that gave teams a two-week head start on day one. We built apps for Amazon, Zoom, Starbucks, Microsoft, GasBuddy, Mercari, and dozens more. Come in, set up the foundations, encode our expertise into tools, accelerate the team, leave them better than we found them.

The tools changed. The job didn’t.

The very first project we ever used Ignite on was a pizza chain app. That’s why the demo is still called Pizza App. Seven years and dozens of client apps later, the message is the same thing Jamon says in every demo: “You can trust the decisions we’ve made here because we’ve shipped them in real apps.”

The handoff

Wax on. Ship off.

We don’t just fix it and leave.

We set up the architecture. We encode our expertise into AI tools, custom skills, and documentation that your team can actually use. We train your engineers — junior devs, senior devs, the ones who just switched from native iOS. We train your Claude, too.

What does that actually look like?

We’ve built skills that automate PR creation so engineers stop burning time on the tedious parts.

We’ve written research skills that pull context from the codebase, relevant docs, and related tickets from Jira or Linear before an engineer writes a single line of code.

On one project we built a testing workflow that ran AI against real human QA checklists — the same checklists the manual testers use — so the team caught regressions before reviewers even opened the PR.

Clients get docs built from real sources — modern articles, client war stories, fundamentals we’ve learned the hard way — not the generic stuff you get from a deep research prompt. The specific linting rules and “this pattern will bite you on Android” notes that only exist because we’ve shipped these apps. Engineers on the project said “I don’t know how we lived without this before.”

The point isn’t any one of these. It’s that when we leave, your team has the setup, the skills, and the muscle memory to keep going without us.

After we leave, your team ships faster than before we showed up.

TL;DR

Believe it or not, it goes deeper.

The barrier to building dropped to zero. The barrier to shipping didn’t.

That’s our lane. Always has been.

Need help getting from prototype to production? Reach out to Infinite Red. We’ve been doing this for over a decade, and we’re better at it now than ever.

Been through this yourself? Come swap war stories at Chain React, the only React Native conference in the US.

Want to start with our tools? Check out Ignite on GitHub.

Special thanks to Steve Connor, Silas Matson, and Jon Major Condon for their ongoing AI research that informed this piece, and to Jamon Holmgren for being everyone’s unpaid app mechanic.