Manga Method

What it was

Manga Method was a language-learning app built on comics. Specifically webtoons, the long comics formatted to scroll on a phone. You read a story in the language you were learning. Tap a line of dialogue to hear a voice actor say it. Double-tap for the full-sentence translation. Tap a single word for that word on its own, translated and pronounced.

The part I cared about was where the translations and audio came from. Rather than produce all of it in-house, we let users contribute their own. That let the app support language pairs a commercial product would never bother with, like Korean straight to Spanish instead of routed through English.

What got built

A working v1: the reading interface, a CMS for me to load and manage comics, and the contribution system. Users came, and they contributed thousands of lines of translated dialogue across several languages. Voice recordings came in too, in smaller numbers.

I didn’t write the code. I hired an agency and ran the build — every screen and system was mine to spec and manage. This was my role before AI tools existed to do the building: product manager, not engineer.

Why I stopped

I did the math. I added up what it would take to get from v1 to a version that could sustain itself — a full-time developer, ongoing content sourcing — against what subscriptions might realistically bring in. It didn’t close. Then ALR sold, which took the traffic I’d have used to acquire users, so I shelved it.

The idea was sound. What killed it was the price of building it the pre-AI way, with an outside agency doing every change.