AnDc builds AI applications for the classroom. They notice how a student is thinking, not only what they answered, and they give teachers back the evenings that marking takes.
In a class of forty, a forty-five minute lesson leaves about a minute for each child. Feedback arrives days late, if it arrives at all, and by then the mistake has set into a habit.
Teachers know this better than anyone. They are not short of care. They are short of hours, and no one has built them a tool that respects how they actually teach.
We design for large classes, shared devices and unreliable connections, because that is where most learning really happens.
Talks a student through a mistake with questions instead of answers. Works in Bahasa Indonesia and English, and stays on the curriculum the school is teaching.
Reads typed and handwritten work and drafts feedback against the teacher's own rubric. The teacher edits and approves. Nothing reaches a student unreviewed.
Shows which idea the whole room is stuck on before the next lesson begins, so the teacher spends the first ten minutes where they matter.
Five layers, one rule: nothing the AI writes reaches a student until a teacher has said yes.
Tools that replace teachers get switched off. Ours are built to make a good teacher's judgment go further, and that is what keeps them open on the desk year after year.
We build for local languages and local syllabi first, not as a translation added later. Student data stays under the school's control, and every AI suggestion can be traced to the work that prompted it.
Schools are buying software again, and almost none of it was made for crowded rooms, mixed devices and patchy connections. That gap is wide, and it is ours to close.
We are opening conversations with investors who care about learning outcomes as much as returns, and who are willing to be patient while trust is earned one school at a time. If that sounds like you, we would like to meet.
Write to us