The URL import workflow: how to mine your reading list for vocabulary

Be honest about your reading habits. You probably read three or four articles a day in your native language, half a book a month, and zero novels in your target language unless you’re actively in a "trying to be a better learner" phase. That ratio is completely normal. It is also the leverage point most language apps miss.

Lemnly’s URL import was built to meet you where you already are. Paste a link, get vocabulary. Three minutes a day adds up faster than you’d guess. This post is the practical guide to running that workflow well.

Why URL imports are the killer feature for most learners

• Articles are sized like attention. A 1,500-word op-ed is roughly the longest thing most people can finish in one sitting. That’s exactly the right size for an import — small enough to process quickly, large enough to yield 10–30 new cards.
• Articles renew themselves. A novel is one decision: "am I reading this book?" A reading list of articles is a hundred small decisions, and the topic variety keeps you interested.
• You’re already doing it. If you read Le Monde or El País or Frankfurter Allgemeine on your commute, the only missing step is pasting the URL into Lemnly when you’re done. Five seconds.

The daily routine

1. Read the article first. Don’t paste it into Lemnly and then read the import preview — that breaks the comprehension flow. Read normally. Tolerate not knowing some words.
2. Paste the URL into Lemnly when you finish. Whether you read it in a browser or a read-it-later app, the URL is what we need. Lemnly extracts the readable text the way Pocket or Reader does — chrome stripped, just the article body.
3. Skim the preview. Lemnly will show you every word it’s proposing as a new card, alongside the sentence it came from. This is the only part of the workflow that needs judgment.
4. Trim aggressively. If a word appeared once in a parenthetical aside and you’ll never need it again — drop it. The cards you add today are the reviews you’ll do for two years.
5. Commit. One button. The cards land in your deck with a low daily-new-card drip so you don’t get a 60-card surge tomorrow morning.

Total time per article: 60–120 seconds. Yield: 5–25 cards. Habit: as cheap as it gets.

How to set up your reading list to compound

A few habits that turn the URL workflow from "occasional thing" to "background process":

• Subscribe to three newsletters in your target language. Newsletters are short by design and arrive on a schedule. That regularity is what you want.
• Bookmark a small set of "evergreen" sources. A long-form magazine, a opinion column, a recipe site, a tech blog — varied vocabulary across registers.
• Use a read-it-later app as a buffer. Save articles when you find them, read them in batches. The URL is still pasteable into Lemnly from the read-it-later app.
• Don’t import what you didn’t finish. Cards from articles you bailed on are cards with no emotional weight. They forget faster.

What the import flow does behind the scenes

A quick tour, for the curious:

1. We fetch the URL and run a readability extractor — same family of algorithm as Mozilla Readability or Postlight Parser.
2. We detect the language from the extracted text. (If we get it wrong, you can override before importing.)
3. We tokenise and lemmatise. "Was reading" becomes the lemma "read," and the past-tense form is remembered as context, not as a separate card.
4. We dedupe against your existing vocabulary and against the shared cache. Most words you "encounter" in an article have been glossed before by another learner — no AI call needed.
5. For the genuinely new lemmas, we call Claude Haiku to generate a gloss, an example sentence (often pulled from the article itself), and a part of speech.
6. We show you the preview. You decide. Nothing is added without your click.

Working with paywalled sites

Some publications gate their articles behind a login. Lemnly can’t bypass that — nor should it. The simplest workaround: when you’re reading the article in your browser, use our extension (Chrome, Firefox, Safari) to import the page you’re looking at. The extension sees what your browser sees, so logged-in content imports the same as anything else.

We don’t store the article body after import. We only keep the lemmas you accept as cards, and the example sentence you chose to attach.

The compounding, one week at a time

A reasonable beginner week looks like this:

• Mon: one article. Import. 12 new cards.
• Tue: daily review (5 min). One article. 8 new cards.
• Wed: daily review. Skip the import — busy day.
• Thu: daily review. One article. 14 new cards.
• Fri: daily review. Skim the week’s preview history. Drop anything that, in hindsight, doesn’t feel worth a card.
• Sat: daily review. Read a longer essay. 22 new cards.
• Sun: off.

Weekly net: about 50–60 new cards. Monthly: 200–250. Yearly: 2,500–3,000 — which puts you at the 95% coverage threshold for most languages in roughly two years, starting from B1.

One more thing

URL imports work because they remove the friction between "I am reading" and "I am studying." If the two activities feel like one thing, you keep doing both. That’s the whole product, in one sentence.