Changelog

I built three full versions of Moglearn before this. None were what I wanted to release. I’m now working on v4 and getting it ready for the MVP.

Planned

Still fixing word segmentation, contextual definitions, names and pronunciation help across the courses. Some generated definitions were using the wrong meaning of a word even though they passed the formatting checks.
Deferred the canonical definition system until after the MVP. For now I’m focusing on getting the meaning right in each reading.
Removed the website sampler, added a language popup, restored the full About page and rebuilt the waitlist page to match the site. Removed public sign-in links and made the footers consistent.

Integrated all eleven course options into the local v4 app: 2,766 readings, including Simplified and Traditional Chinese.
Finished the editorial review and a follow-up correction pass. Corrected errors in 57 readings and 218 practice questions, including Traditional Chinese conversion mistakes.
Tested course navigation, language switching and saved drafts on desktop and mobile.
Compared Deepgram Nova-3, Nari and browser Zipformer on the same 56 recordings, including accented English. This was a small test, not validation for every language.
Decided to use prerecorded partner audio for the MVP. The recordings still need to be made.

Connected the reader, listening and practice flows. Added resume and skip handling so they don’t interrupt an answer already in progress.
Fixed draft restoration across reloads and language changes. Earlier answers stay saved when I revise a question.
Revised practice prompts, examples and retry guidance. Added more chances to ask questions and reuse language from earlier readings.

Deployed the email-only waitlist and ASCII website to moglearn.com.
Set up launch-update email delivery and unsubscribe handling. No launch email sent yet.

Wired the new account screens into the local v4 app. They had been built in a separate checkout and weren’t connected to the main preview.
Tested sign-in, sign-out, account errors and returning to the course, including phone layouts.

Started v4 with a fresh frontend.
Narrowed the MVP to the course: reading, word help, listening, speaking and writing practice. Cut the media importer, broad content library and elaborate dashboards from the first release.
Kept the speech and storage code that was still useful, along with the existing content and saved work.

Built and fixed reader controls, listening playback and saved practice in the earlier app.
Reworked the English and Chinese beginner lessons and corrected practice questions that didn’t match their readings.
Rejected generated definition batches after review found incorrect meanings.
Split the reader, admin screens and backend Worker into smaller modules.

Moved application data and server work from Firebase and Cloud Run to Cloudflare.
Built out the reader and tools for preparing readings, definitions and audio.