XamKids
Built at Meta Global AI Developer Hackathon · Oct 26, 2026 · Remote

In Senegal, most children learn in French, which is often not their first language, and many struggle to read and understand their lessons, especially those who learn differently (dyslexia, attention or number difficulties). Teachers notice it, but have neither the time nor the tools to adapt every lesson for every child. XamKids is an AI agent that turns the teacher's own printed lesson into lessons every child can understand. The teacher photographs a textbook page. The agent reads it, identifies the key concepts and prepares an easy-reading version, a step-by-step focus version and a simplified version, with an interactive visual explanation and comprehension questions. The agent checks its own work before showing it to the teacher: every version must keep all the key concepts of the original, and every visual is inspected and redone if something is wrong or missing. The teacher can ask for corrections in plain language, such as "the second paragraph is too long", and the agent changes only what was asked. Nothing reaches students until the teacher validates it. Each child then enters their own space and gets the validated lesson adapted to their level and interests. They can listen to it, use reading supports, and answer the questions by tapping, writing or speaking. The teacher follows the progress of her class. XamKids does not diagnose children, does not chat freely with them, and keeps minimal data. It is built for the reality of Senegalese public classrooms: large classes, teachers' existing lessons, and children who need the same lesson in a different form.