Matic Cues: On-Device Voice and Gesture Control for Robot Vacuums
Published by AINave Editorial • Reviewed by Ramit
Matic has released a software update called Matic Cues that lets you control its robot vacuum with voice commands in over 70 languages or by pointing at a mess. The feature runs entirely on-device unless you opt into cloud processing, making it a notable example of privacy-preserving natural language interaction in home robotics. For AI builders, the update demonstrates how on-device multimodal AI can replace app-based control in consumer hardware.
Voice and gesture control without the cloud
Matic Cues is powered by Google Gemini and responds to spoken commands like "Hey Matic, vacuum the kitchen" or "clean underneath the cabinets" without requiring a phone or app. The robot understands variations in phrasing and supports commands across rooms or the whole home. Gesture control works by pointing at an area and saying "clean this" -- the robot combines camera input with voice to locate the spot and clean it. Both modes process data on the robot itself; cloud sharing is opt-in only.
The update is rolling out over the air to all existing Matic units and is included on new robots. Matic costs $1,245, and Cues is free for current owners. The company also added Apple Home and Siri support via Matter earlier this year, along with Live Activities for real-time status on iPhone, Mac, and Apple Watch.
Why this matters for AI builders
Matic Cues is a practical case study in on-device multimodal AI. The robot uses five cameras and a vision system built by former Nest engineers to understand rooms in 3D, then combines that with speech recognition to act on natural commands. For builders shipping AI products, the key takeaway is that privacy-first architecture doesn't have to sacrifice capability -- the robot handles voice, vision, and gesture locally while still offering cloud opt-in for users who want it.
The integration with Matter and Apple Home also shows how consumer robotics can slot into existing smart home ecosystems without requiring a proprietary app for every interaction. That matters for adoption, especially in households where not everyone is comfortable with a phone-based interface.
What remains unclear
The source evidence for this update comes from a single hands-on review, so independent verification of the claimed 70+ language support and on-device processing guarantees is limited. Gesture control reportedly takes slightly longer than voice-only commands, and the reviewer notes that the app's Draw to Clean feature is often more precise for spot cleaning. Cloud opt-in is available but the privacy implications of that choice are not detailed. Long-term reliability data for the on-device AI stack is also absent.
For builders evaluating similar approaches, the trade-off between on-device latency and cloud flexibility is worth watching. Matic Cues shows that a well-designed on-device pipeline can feel natural, but the real test will be how it handles edge cases and updates over time.
