Android XR Launches With Gemini 3.5 Wearable Agent Support

Google marked its return to the wearable computing market at I/O 2026 with a dual-hardware strategy for Android XR. The spatial computing operating system powers a new category of “Intelligent Eyewear,” deliberately splitting the ecosystem into an immediate audio-only hardware release and a longer-term visual display track. By decoupling the form factors, Google aims to sidestep the thermal and battery constraints that have historically blocked continuous wearable AI.

Two-Tier Hardware Ecosystem

Google is pursuing two distinct form factors to scale Android XR across different consumer thresholds. The first tier, confirmed for a Fall 2026 release, consists of audio-only glasses developed in partnership with Samsung and Qualcomm. These devices lack an internal display entirely. Instead, they rely on built-in cameras, microphones, and speakers to facilitate voice-based assistance. To avoid previous market stigmas regarding wearable camera designs, Google partnered with Warby Parker and Gentle Monster for the physical frames.

The second tier is a display-equipped prototype internally named Project Aura. The hardware features a single-window color waveguide display over the right lens, offering a 20-degree field of view with an embedded microprojector. Google has not committed to a shipping timeline for the display variant, explicitly citing unresolved battery life limitations. A separate teased collaboration with XREAL showcased a tethered version of Project Aura featuring a wider 70-degree field of view using OLED technology.

Agentic Workflows and Multimodal AI

Both hardware tiers are driven by Gemini 3.5 Pro and Gemini 3.5 Flash, which handle local scene understanding and task routing. Processing is heavily offloaded to a companion smartphone—compatible with both Android and iOS—to keep the eyewear lightweight.

Demonstrations at the event highlighted the system’s focus on autonomous task execution rather than simple voice search. Using Gemini Spark, the glasses execute multi-step routines while the companion phone remains in the user’s pocket. In one validated test, the system handled a “look-and-ask” multimodal request to order a specific coffee, seamlessly navigating the user to the shop while autonomously driving the DoorDash interface on the tethered phone.

The audio-only frames feature a real-time translation capability equipped with “voice-matching,” which mimics the original speaker’s pitch and tone. The display prototype layers visual data over this experience, projecting live translation captions, turn-by-turn walking directions, and real-time application states directly into the user’s line of sight.

Android XR Developer Path

The introduction of Android XR signals a shift in how developers will target wearable AI agents. Google launched an early developer program focused on building “AI widgets” tailored for the Heads-Up Display (HUD).

The ecosystem extends across Google’s existing hardware stack. Photos captured through the glasses can be instantly viewed or edited on a smartphone or Google Watch using the Nano Banana image editing model. Analysts note this OS-level integration presents a direct challenge to Meta’s dominant position in the smart glasses market by offering a clear migration path from audio-only interactions to full visual overlays.

If you build multimodal applications, the immediate hardware reality dictates an audio-first approach. Developers should optimize agentic workflows for voice input and headless execution now, treating the eventual visual HUD as a supplementary canvas rather than the primary interface.