About
We are building the structure Chinese learning ignores.
The starting point
Chinese365 began with a single observation: most apps in this space teach about Chinese, but none of them teach Chinese as it actually is. Vocabulary lists, flashcard decks, lesson pipelines — these all assume the reader absorbs the language through repetition. They work, but they throw away what makes the language interesting: Chinese is a structured system, not a 5,000-word pile.
Every character is built from radicals and components. Every word tags with part of speech and HSK level. Every grammar point comes with examples and typical positions. We expose this structure — first as a knowledge base, then as tools that read from it, then as an API, then as a daily journey app.
What we are not
- • Not a Duolingo clone. We are not building lessons; we are building primitives.
- • Not a dictionary. Pleco already does that.
- • Not an LLM wrapper. We use LLM only where it earns its keep — generation and feedback. The data drives every answer.
The moat
Three long-term advantages compound over time:
- 1. A structured knowledge base over free-form scraping. Every entry is hand-curated or CC-permissioned.
- 2. A per-user knowledge graph. Every mistake you make, every mastery tick, every reflection score — those become signal only we have.
- 3. A proprietary error dataset. Tones, measure words, 了 vs 的, word order — the failure modes of native English speakers writing Chinese. Built from anonymised, consent-first user data.