What langweave is
langweave keeps track of how much of a target language you can read, so your assistant can write part of every reply in it — single words at first, whole sentences as your vocabulary grows. You do not study in it and it schedules nothing. You keep using your assistant the way you already do, and the language turns up in the replies at the level you can actually follow.
It is an MCP server: a connector you add to your assistant once. Today it teaches Spanish and Japanese, to English speakers.
What a conversation looks like
Section titled “What a conversation looks like”Before your assistant answers, it fetches your profile — which words you know, which you are learning, which few to introduce next, and the limits it has to stay inside. It writes the reply with that profile in hand: words you know appear as they are, words you are learning carry a short gloss in parentheses the first time, and at most a couple of new words are introduced per reply. Then it logs every target-language word it used, and that log is how a word moves from new to known.
Day one, a reply about planning a weekend might carry one Spanish word. A few months in, the everyday nouns and verbs are Spanish and only the rare ones are not.
What it is made of
Section titled “What it is made of”- The connector at
https://api.langweave.tech/mcp— five tools your assistant calls on its own, two of them on the Free plan (Plans). You sign in once with OAuth; there is no API key. Connect it → - The learner app at app.langweave.tech — where you create your account, choose a language, look at your progress, add words you already know and change the settings.
- Your memory, kept by lexmem — your account, the word list per language (built from open dictionaries and frequency lists, in frequency order, so the first words you meet are the ones you will meet most) and which of those words you know. langweave reads it to write the profile and, on Standard, records what was woven into it.
There is no level and no placement test. A new learner starts empty, and the profile grows from what is woven into your replies and what you tell it.
Where to go next
Section titled “Where to go next”- Quickstart — from nothing to a woven reply in about five minutes.
- How a word becomes known — the rules, with the numbers.
- The source — the server, the learner app and how to add a language.