Why I Built a Custom Curriculum Feature for Chamelingo

The Bookshelf Problem

I have three Korean textbooks on my shelf. Seoul National University's Korean, Yonsei Korean, and a TOPIK prep book. They've been sitting there for months, and I barely open them.

It's not that the books are bad. They're great. The problem is that opening a textbook feels like work. There's no feedback loop, no streak to protect, no reason to keep going other than willpower. And willpower runs out.

So I did what most people do. I downloaded Duolingo. I did a few lessons. I learned how to say "the cat is on the table" in Korean. Useful? Maybe. But it had nothing to do with what my textbooks were covering. I was studying two completely different curricula, and neither of them stuck.

That's when the frustration really hit. I had perfectly good study material sitting on my shelf, and no tool that could make it feel like anything other than homework.

Why Nothing Out There Worked

I tried everything. Here's what I found:

Duolingo has great gamification, but the curriculum is generic. Everyone follows the same path. You can't import anything. If you're studying with a specific textbook or preparing for TOPIK, Duolingo doesn't care. It's going to teach you what it wants to teach you.

Anki is the opposite. It's incredibly flexible. You can import anything, build custom decks, and it has real spaced repetition. But using Anki feels like doing flashcards in a basement. There's no gamification, no exercise variety, no reason to come back other than guilt. I'd build a deck, use it for a week, and then abandon it.

Textbook companion apps are just digital versions of the same boring experience. Read, memorize, repeat. No interaction, no adaptation, no fun.

What I actually wanted didn't exist. I wanted something that could take my textbook vocabulary and turn it into a game. Not a simple flashcard game. A real training system with different exercise types, spaced repetition, progress tracking, and maybe even a way to compete with other learners.

What I Actually Wanted to Build

The idea was simple, even if building it wasn't:

• Import my textbook vocab and get 23+ exercise types generated from it. Not just "what does this word mean?" but listening exercises, writing practice, sentence construction, fill-in-the-blank, multiple choice, matching, speed rounds, and more.
• Spaced repetition that adapts to me. Not a fixed schedule, but an algorithm (we use FSRS) that learns which words I struggle with and surfaces them more often.
• Battle other learners using my own content. The Arena is one of my favorite parts of Chamelingo. You study your material, then you fight someone else in a real-time quiz battle. It doesn't matter if you're studying different textbooks. The system generates questions from both players' content.
• Track what actually matters. Not generic XP. I wanted to see how well I knew the vocabulary from Chapter 3 of my textbook. Which words I keep getting wrong. What my weak spots are.

Building It Was Harder Than I Expected

The hardest part wasn't the code. It was figuring out what actually helps people learn versus what just sounds cool.

Early on, I built exercise types that felt fun but didn't teach anything. Tapping colorful buttons is satisfying, but if the learner isn't forced to recall the answer from memory, the exercise is basically useless. I went through dozens of iterations, testing each exercise type against what research says about memory retention.

The import system was another challenge. Some people want to paste a list of words. Others have Anki decks with thousands of cards. Some want to upload a CSV from a spreadsheet. Making the import flexible enough to handle all of these without being confusing took a lot of trial and error.

And then there was the balance between the built-in curriculum and custom content. Chamelingo ships with a full 16-chapter grammar course and 300+ vocabulary items. I didn't want custom imports to feel like a separate, second-class feature. Custom content gets the same exercise types, the same spaced repetition, the same Arena battles, and the same analytics as the built-in curriculum. It's all one system.

How It Works Today

Here's the short version:

1. Import your material. Upload vocabulary from Anki decks, CSV files, or just type words in manually. You give us Korean + English (and optionally romanization, example sentences, tags), and we handle the rest.

2. Exercises are auto-generated. From your vocabulary, Chamelingo creates 23+ exercise types. Listening comprehension, writing from dictation, sentence building, conjugation drills, visual matching, timed challenges. The system knows which exercise types work best for which words.

3. Study and compete. Your imported content goes through FSRS spaced repetition. You review when the algorithm says you should. And when you're ready, take your vocabulary into the Arena and battle other learners in real-time.

For the full breakdown with a comparison table, check out the Custom Curriculum page.

What's Next

This feature is live, and people are using it. But there's more I want to do. Better textbook parsing (imagine pointing your phone camera at a textbook page), collaborative decks that study groups can share, and deeper analytics that tell you exactly where your weak spots are.

If you've been looking for a way to actually use those textbooks collecting dust on your shelf, give it a try. Import your first deck and see what happens. And if something's missing or broken, tell me. I built this because I needed it, and I'm building the rest based on what learners actually need.