About Lemnly
One person, one stubborn problem.
Lemnly is built by me — Sebastian Walter — out of Sooties LLC in Wyoming, USA. It started as a Sunday-afternoon problem and quietly turned into a product. There’s no team yet. There’s no funding round. There’s just one person who got tired of forgetting Italian words.
The premise
Every language learner eventually hits the same wall. The first 2,000 words come from courses, the next 3,000 come from willpower, and the 5,000 after that come from doing the one thing the courses can’t do for you: reading a lot of real text.
The blocker isn’t will. The blocker is friction. Books drown you in unknown words. Looking them up breaks reading. Building cards by hand is glacial. Most learners abandon the project somewhere around chapter four.
What Lemnly does today
A single pipeline: text in, vocabulary out. Paste a URL, drop an EPUB or a DOCX or a plain text file, snap a photo of a page, or speak a word. Lemnly tokenises and lemmatises what you give it, checks it against a shared cache and your own history, and proposes only the cards that are new and at your level. The scheduler is FSRS, with separate forward and reverse review for each card. There are five study modes — flashcard, typing, listening, multiple-choice, letter jumble.
What it doesn’t do yet (and what’s coming)
- Languages. Today: English, German, Spanish. French, Italian and Portuguese are next.
- Native apps. The codebase is Expo, so iOS and Android builds are on the path. The web app installs as a PWA in the meantime.
- Browser extension. On the list. For now, paste the URL.
- Paid plans. The beta is free. When paid plans launch, existing users will get notice first.
What I believe
- Software should help you read more books. If an app keeps you in its own UI instead of pushing you toward the book you wanted to read, it failed.
- Streaks should encourage, not punish. A missed Tuesday is not a moral failure.
- AI is a labelling tool, not a teacher. The learner does the work. The model saves time on the tedious parts.
- Your data is yours. Export at any time. Nothing is sold; nothing is used to train models.
Get in touch
If you’ve got feedback, a feature request, or you just want to talk about how to read Tolstoy in the original — please say hello or email hello@sooties.com. I read every message.
The article you’d normally skim?
Paste it in tonight.
Free while in beta. Add your first source in under a minute and see the words you actually need.