Local-First Software

A software design philosophy, articulated by Kleppmann and colleagues in 2019, in which applications keep the user's primary data on local devices and treat cloud services as optional synchronization or backup layers rather than as the source of truth. Local-first systems aim to give users ownership of their data, offline functionality, low latency, long-term durability independent of any provider, and stronger privacy guarantees. The approach is increasingly relevant to accessibility because on-device AI models (speech recognition, image description, text simplification) can process sensitive environmental data — speech, faces, locations — without transmitting it to third-party servers, and can deliver the sub-second latency that live captioning and real-time interaction require.