In Addis Ababa, Ethiopia, internet access comes with what locals call the "Internet Tax" - data costs that put cloud-based AI tools out of reach for most of the country's 35 million students. Ivy is a new AI tutor built specifically for this reality.
Runs on a $120 Phone, No Internet Required
Ivy uses edge inference - meaning the AI model runs directly on the device instead of sending data to a server. It works in full airplane mode on smartphones costing as little as $120. Quizzes and flashcards render as interactive widgets inside the chat interface rather than requiring separate app screens or menu navigation.
The more unusual feature is that Ivy is proactive. Most AI tutoring tools wait for the student to open the app and ask a question. Ivy initiates outbound voice calls to check on students who miss scheduled sessions. It's a design choice that mirrors what human tutors actually do - follow up when someone falls behind.
A Different Design Philosophy
Most AI education tools assume reliable broadband and modern devices. Ivy flips those assumptions entirely. The offline-first approach means no streaming tokens from a cloud API, no latency issues, and no recurring data costs for the student.
The tradeoff is obvious: a model small enough to run on a budget smartphone won't match GPT-5 or Claude on complex reasoning tasks. But for structured tutoring - drilling vocabulary, practicing math problems, reviewing concepts - a smaller model that's always available beats a powerful one that's too expensive to access.
Ivy is currently focused on the Ethiopian market, but the underlying approach applies anywhere internet access is unreliable or expensive. Rural areas, developing nations, even subway commutes where you lose signal. Offline-first AI is an underserved category, and Ivy is one of the few projects treating it as a feature rather than a limitation.