Kwame 2.0, a bilingual AI teaching assistant, successfully supports 3,717 coding students across 35 African countries through a human-in-the-loop approach.

Kwame 2.0 is a generative AI teaching assistant built with retrieval-augmented generation, supporting English and French languages

Deployed in SuaCode, a mobile-based introductory coding course, across 15 cohorts over 15 months with 3,717 enrollments

System achieves high accuracy on curriculum-related questions while human facilitators and community members provide oversight to catch and correct errors

Operates in a human-in-the-loop forum model that encourages peer participation and community learning alongside AI support