Where we build. Only what works stays here.
The Lab has two groups. One is products built for Hong Kong education — mostly free, aimed at newly arrived families and DSE candidates. The other is tooling we built for ourselves and use daily, made publicly available. The second group has nothing to do with the education line, and we won’t dress it up as if it did.
Education products
Built for Hong Kong on the ground. The three form one line: can’t read traditional characters on arrival → can’t follow Cantonese → don’t know how the DSE works.
100 Days of Traditional Characters
A hundred AI-selected worksheets to help newly arrived families write and read the characters around them.
The character list is ordered by everyday frequency in Hong Kong, not by stroke count. Given away free, with no details to hand over.
Position in the funnel Step one · just arrived, the child can’t read traditional characters
Cantonese Scenario Practice
A Cantonese practice partner for newly arrived students: real situations, usable the moment you speak.
Scenarios come from situations users actually submitted — ordering food, asking directions, explaining to a teacher why you were late. One standard of judgement: could the other person understand you.
Position in the funnel Step two · understands it, won’t speak it yet
HKDSE English Chunk Reading
Re-understand English through chunks rather than single words, practised directly against DSE Paper 1.
Reading ability broken into diagnosable sub-skills, then set against past papers. The method itself is free; the accompanying practice is paid.
Position in the funnel Step three · the DSE comes into view
These three were built for newly arrived families from the start — the character worksheets literally open with “the first barrier for anyone new to Hong Kong”. They simply were never organised into a line before. One step further on is the WhatsApp AI Teaching Assistant, which is where a parent first meets an institution-grade system.
OPEN PROBLEM WALLYour problem can go up here, and we answer in public.
That invitation above used to lead only to a private email. Now the problem goes on a public wall and the reply sits on the same page — problems raised repeatedly become the next Lab project.
Go to the wall →Tools we use ourselves
Built internally, publicly available.Unrelated to the education line. Listed here to make one point: we run our own work on AI tools we built.
Send her your voice notes one at a time and she handles the rest — grouping, tidying, reminding, reworking, and a few thoughtful touches you wouldn’t expect.
Our internal task manager, publicly available. It was not designed for students — which is exactly how it differs from SLOT: SLOT’s tasks come from a learning diagnosis, GTD’s come from you.
BUILT INTERNALLY
Put it in real hands first, then decide whether it deserves to continue.
Lab output doesn’t become an institutional solution directly. Its job is to let us make our mistakes on our own products rather than in a client’s classroom.
Build the smallest version first
One PDF or one conversation flow is enough. Build until a real person can use it, not until it’s a finished product.
Real users, not internal testing
How parents and students actually behave is a different thing entirely from us trying it ourselves.
Define failure before starting
It is far too easy to find reasons to keep a project afterwards. The condition for stopping gets set up front.
Only proven mechanisms enter the product line
Nothing reaches the three-layer system unless real users validated it here first.
Got a teaching problem worth trying together?
If you run an institution or a school, or you teach, and something keeps recurring that nobody has solved — bring it. If it fits, we build it as a Lab project together.
Want the official changes in Hong Kong education, daily? Subscribe to MNT Insights →