A voice-first AI assistant with its own personality, built for desktop.
Athena is a voice-first personal assistant, built around a core "brain" that handles intent routing, conversation, memory, and mode control, with pluggable modules — Spotify, calendar, PC control, and more — that can each fail independently without taking the assistant down. Rather than a generic command-response bot, it's designed with a distinct, personality-driven character, and supports both English and Arabic.
Real-time voice pipelines and practical LLM integration are a different discipline from wrapping a chat window around an API — speech recognition, text-to-speech, and local model inference all have to work together with low enough latency to feel like a conversation. Building Athena is us developing that capability directly, module by module, rather than treating it as a black box.
Pre-alpha, in active development.