Why we fund small, task-specific AI instead of general-purpose models, and how that keeps costs low enough to run offline on a basic phone.

Why we care more about getting AI adopted at scale than building it, and what that takes from the governments and partners we fund.