Show HN: RunAnwhere – Faster AI Inference on Apple Silicon
Summary
RCLI is an on-device voice AI suite for macOS that runs a complete STT + LLM + TTS pipeline locally on Apple Silicon. It offers 43 macOS actions via voice, local RAG over documents, and a MetalRT GPU engine designed for sub-200ms latency with no cloud dependencies. The project supports multiple models and includes an architecture designed for live model hot-swaps, with a MIT license for RCLI and a proprietary license for MetalRT.