Our vision

The problem we started from

In our class, PE is two periods a week and English is something you pay extra for. A classmate who cannot afford extra lessons falls behind on both, and nobody notices because neither is measured day to day. What is not measured does not improve.

Three principles we hold to

1. It must run on what people already have

No wearables, no gym, no paid subscription, no app-store install. A browser and a camera, on a phone that is several years old, has to be enough — otherwise the students who need it most are the ones left out.

2. The camera stays on your device

Video frames are never uploaded. The pose model is downloaded into the browser and runs there; the server only ever receives numbers — which exercise, how many repetitions, how long. A fitness app that watches children has to earn trust by design, not by promise.

3. The numbers must be honest

We refuse to invent users. The leaderboard counts real people only; the practice companions that keep an empty board from looking dead are marked 🤖, sit in a separate "Community" class, and are excluded from every statistic. A number nobody can trust is worse than no number.

Where we are going

• More content. Toward 4,000 vocabulary entries and around 125 readings, B1 through C2, across science, technology, economics, history, environment and careers.
• Mock entrance exams. A timed English paper in the format of Hanoi's specialised-school entrance test.
• AI that teaches, not just marks. Written work already gets specific corrections with reasons; next is feedback that adapts to the mistakes a particular student keeps making.
• Other schools. The whole system is one Node.js process and one SQLite file. Any school can run its own copy — the deployment guide is public.

What we will not do

• Sell, rent or share student data.
• Show advertising to children.
• Upload camera footage anywhere.
• Publish fake users, fake reviews or fake statistics.