I built an open-source iOS app that turns Meta Ray-Ban glasses into an on-device AI assistant (neural voice + model picker)
Hey all — I've been building OpenVision, a free and open-source iOS app that connects Meta Ray-Ban smart glasses to AI. Just shipped v2.3.0 and wanted to share it here.
What it does:
• 5 AI backends — fully on-device (Apple MLX + Apple Intelligence, works offline and private) or cloud (OpenAI / Gemini Live)
• Pick your local model — Qwen 2.5, Gemma, or SmolVLM, so you can trade memory for capability
• On-device neural voice (Kokoro) — natural text-to-speech, nothing leaves the phone
• On-device face recognition — "remember this person as Sara", then later "who is this?"
• Live web search that returns current info (news, weather, prices)
• Wake word ("Ok Vision"), conversation memory, fully hands-free
It's MIT-licensed and on GitHub: https://github.com/rayl15/OpenVision
Would love feedback from people who actually own the glasses — especially which models or features you'd want next.