Show HN: We've built a standalone Apple Watch app running LLMs offline, locally (nobodywho.ooo)

A groundbreaking development has emerged with the launch of two standalone Apple Watch apps by NobodyWho, designed to run AI models locally and offline. This innovation marks the first of its kind, allowing users to access personal AI assistants directly on their devices without relying on an iOS companion app. The app aims to offer quick answers to simple questions without persisting conversations, which is ideal for users needing information on the go. It currently utilizes smaller models, such as Prism ML’s LFM2.5 and Liquid AI's 1-bit Bonsai, optimized for the limited computing power of the Apple Watch. The significance for the AI/ML community lies in the push towards more efficient, offline-capable applications that enhance accessibility and user experience. By leveraging the NobodyWho inference engine, the apps demonstrate that even with reduced computational resources, intelligent and responsive AI interactions can be achieved. As smaller language models continue to evolve, this development signals a promising direction for the integration of AI into wearable technology, paving the way for smarter, more capable personal assistants on edge devices.