Skip to content
AppDues
Log in
Contact salesSign up

From Sketch Artboards to Shipped Apps: Matt Birchler's Indie Path

Matt Birchler built a following as a writer, YouTuber, and podcaster long before he shipped anything to the App Store. The turning point was not a business plan. It was a repetitive design task that made him ask why he was still doing it by hand. This guide follows that thread: the chore that became Quick Reviews, the clumsy AI assistance that forced him to actually learn SwiftUI, the reasoning behind charging from launch day, and the large pile of experiments that never became products.

A design chore that turned into a product

The original workflow lived in Sketch. To post a review of a film or game, Birchler opened a document on his Mac, copied the artboard, replaced the poster, and retyped the text. It worked, but it was manual and tied to a desktop.

When readers asked which app produced those graphics, he published a one-page jQuery tool that automated the layout. That version still ended with a screenshot, and it had no accounts and no saved history. The obvious next step — using the tool on a phone — required native iOS skills he did not yet have.

A rough Swift weather app already sat unfinished in his App Store Connect account. Quick Reviews became the first release he treats as serious, and it set the pattern for what followed: notice a repeated irritation, then test whether a small tool can remove it.

Why weak AI help was useful help

Birchler already knew HTML and CSS. He has described his earlier iOS attempt as feeling like laying out a page in tables, so SwiftUI's declarative style felt closer to what he understood. Early assistants such as Cursor were arriving at the same time, and he spent roughly six weeks on the app.

The tools were crude. He recalls having to request one small change at a time, watching it break, and repeating the request — and having to remind the assistant that a single Swift file was not a sensible structure. That friction had a side effect: to get anywhere, he had to read the generated code, follow the data structures, and understand how views connected.

That is the distinction worth keeping. Letting a model produce code you never read gets an app built once. Reading it builds the judgment to direct better tools later and to check whether their output is actually correct.

Charging from day one to cover dependencies

Quick Reviews is free to use, with a $10 annual subscription available from the first release. The motive was not an aggressive growth target. It was exposure to services he does not control.

The app draws metadata, posters, and cast details from The Movie Database and imports recent Letterboxd reviews through an RSS feed. Any of those could change terms or start charging. Putting the automated metadata and the Letterboxd sync behind the paywall gave the app a way to fund those costs rather than absorb them.

His stated aim was simple: the projects should not lose him money. Across the wider catalog, the apps have become his largest side-income stream, ahead of both his blog and his YouTube channel by a wide margin. Treat that as one developer's account rather than a benchmark for what a small app earns.

Small scope, clear names, many unfinished folders

Once the habit formed, more annoyances became candidates. Quick Subtitles and Chapterize came out of his own podcast editing routine and get used several times a week. When he set a goal of running 365 miles in a year and could not find a workout app that simply charted progress toward that number, he built Yearly Run Goals himself.

Naming follows the same logic. Words like "Quick" tell a user this is a focused utility rather than a sprawling platform with endless support obligations. Scope is a promise about maintenance as much as about features.

Not everything ships. A folder named "Apps" in his home directory holds 99 folders — a mix of his own projects, small tools, abandoned visualizers, and ideas that stopped early. Some friction points turn out to be fixable; others do not. Keeping the failures costs almost nothing and preserves the option to revisit them.

A pre-build checklist for the same approach

Before starting a small utility of your own, answer these in writing:

  1. What exact repeated action are you doing by hand, and how often? If it happens once a quarter, the tool probably will not earn its maintenance.
  2. Which parts depend on someone else's service, and what happens if that service adds a fee or changes access?
  3. Which features are automated and which are manual? The automated ones are the natural candidates for a paid tier.
  4. Can you describe the app in a name a stranger understands in one second?
  5. What will you deliberately not support, so the scope stays honest?
  6. If the first version fails, where does it go — and what would make you reopen it?

Worked example: the running goal was 365 miles in a year, and the missing feature was a simple graph of progress toward that number. The scope decision was to build the graph, not a full training platform. That single constraint is what kept the project small enough to finish.

What the full conversation covers

The complete episode goes further than this summary. Birchler explains how actually selling software through the App Store changed his view of Apple's 30% commission, why YouTube sponsorships can generate more administrative work than they are worth, and how co-hosting the Comfort Zone podcast gives him a collaborative creative outlet.

Friction-to-app map
Friction noticedWhat was builtScope decision
Posting movie and game reviews required manual Sketch edits on a MacQuick ReviewsAutomate metadata and Letterboxd import; keep the rest free
Podcast production steps repeated weeklyQuick Subtitles and ChapterizeNarrow tools used several times a week
No workout app charted progress toward a 365-mile yearly goalYearly Run GoalsGraph the target instead of building a training platform
Ideas that never resolved into a shippable toolUnreleased experiments in the Apps folderStop early and keep the folder

Descriptions reflect one developer's reported experience; the folder count and app list are as described at the time of the conversation and may have changed.

Dependency and paywall boundary
ComponentSource of dataCovered by paid tier?
Poster, cast, and title metadataThe Movie DatabaseYes — automated metadata sits behind the paywall
Recent review importLetterboxd RSS feedYes — sync sits behind the paywall
Manual review compositionOn-deviceNo — free to use

Boundary as described for the launch configuration; feature placement can change between releases.

Frequently asked questions

Did Matt Birchler start as a developer?

No. He came from television production study, retail management, and tech support, and worked as a blogger, YouTuber, and podcaster. An unfinished Swift weather app marked his first attempt; Quick Reviews is the release he counts as his serious start.

Why does a free app have a $10 annual subscription?

Because the app relies on outside services. Metadata and artwork come from The Movie Database, and recent reviews are imported from a Letterboxd RSS feed. Charging for the automated parts from launch gave the app a way to cover those costs if a dependency stopped being free.

What did early AI coding tools actually teach him?

Structure. Because the assistant kept breaking small requests and cramming everything into one file, he had to read the output and learn how data and views fit together. That knowledge is what lets him direct stronger tools today and judge whether their code is right.

Do all of his app ideas get released?

No. The Apps folder in his home directory contains 99 folders, including his own projects, tiny tools, abandoned visualizers, and experiments that stopped early. Some friction points turn out to be fixable and ship; others stay in the folder.

Are the apps really his biggest income source?

He reports that the app catalog brings in more than his blog or his YouTube channel by a significant margin. That is a personal account from one developer, not a published benchmark, and figures are not disclosed.