Saymore is a cross-platform desktop voice input tool positioned as a local-first dictation and text refinement solution. It uses a global hotkey to invoke recording, converts speech to text, and inserts the result directly at the current cursor position, working across editors, browsers, chat apps, and other scenarios. Its core mechanism is a modular provider architecture: both speech recognition (ASR) and text refinement (LLM) can be configured to use local models or cloud services, allowing users to freely combine them to balance privacy and performance. Built with Rust and Slint, it emphasizes data sovereignty, avoids vendor lock-in, and supports encrypted history with configurable retention policies.