Meta's Muse AI agent reaches its smart glasses with a wave of accessibility and navigation updates

Meta's new AI agent, Muse, is rolling out to the company's smart glasses — including the models unveiled at Meta Connect — just two weeks after its initial launch. Users will be able to invoke the agent by name and ask it to run a fitness session, log meals, or help purchase products they are looking at. At the same time, Meta is releasing a bundle of new capabilities spanning accessibility, imaging, audio, and navigation, signalling a significant push to make the glasses a more standalone device rather than a phone accessory.

Muse on your face: voice activation and daily tasks

According to Meta, Muse will enable more natural interaction with the glasses without pulling out a smartphone. The agent is designed to work as a contextual assistant: when a user asks for workout guidance, the system uses sensors and the camera to track movements; when they want to log a meal, the camera photographs the dish and the AI identifies it and records nutritional values; and while shopping, the glasses recognise the product in view and surface information or a purchase option. The company has not published performance metrics or latency figures, so real-world behaviour outside demos remains to be seen.

Accessibility: FDA-cleared hearing enhancement and richer visual context

The most prominent accessibility update is Hearing Enhancement, software that has received FDA clearance for adults with mild-to-moderate hearing loss. The feature will be available later this year through a Meta One subscription or as a $149 add-on for compatible glasses. Separately, Meta AI is slated to deliver "richer contextual" answers when users ask what they are looking at, a capability intended to help blind or low-vision users interpret their surroundings. A new audio-balance feature will let wearers shift volume between ears, including routing all audio to one side — useful for people with asymmetric hearing.

Imaging and audio: Dolby Atmos capture and dynamic frames

On the media side, the glasses gain video recording with 360-degree spatial audio via Dolby Atmos Capture, a first for this product category. A Dynamic Photo feature automatically captures a burst of frames with every shutter press, then Meta AI selects the best frame from each cluster; the concept resembles Apple's Live Photos but with automatic instead of manual selection. Meta has not disclosed which vision model performs the selection, nor whether processing occurs on-device or in the cloud.

Ray-Ban Display: holographic avatars, navigation, and on-screen editing

The Ray-Ban Display models receive the "hologram" feature unveiled alongside the new VR headsets: a photorealistic version of the user's face for video calls that "infers" expressions from voice alone. The feature will launch in early access on WhatsApp for Display users in the United States this autumn. For navigation, starting next month users in the United States and Canada will be able to request voice-guided cycling or public-transit directions in real time, while walking directions will be guided by landmarks — home, car, sign — rather than street names. Also on the way: on-glasses photo editing, an Explore app for app discovery, an improved calendar, and phone notification mirroring.