How contributions work
From a speaker’s answer to a learner’s lesson, every step is traceable.
1. A guided conversation
The AI interviewer asks one plain-language question at a time — “How do you say ‘water’?”, “Can you use it in a sentence?”. You can type, record, upload audio, or say “I don’t know”.
2. You check the summary
Vault shows what it understood, clearly labelled as a suggestion. You correct anything wrong. Your original answers are stored exactly as you gave them.
3. Automatic checks
Before people review it, Vault looks for possible duplicates, other speakers’ answers to the same question, and missing details like a recording. These are hints for reviewers, not decisions.
4. Human review
Reviewers approve, ask you a question, accept your answer as a variant, or open a discussion if answers conflict. Disagreements are settled by an admin — and every form is kept.
5. Publication
An admin publishes reviewed entries. Words need a native-speaker recording that you agreed learners may hear.
6. Lessons
Lessons are drafted only from published entries, then approved by an expert before learners see them.