Smart Glasses & AI Wearables

Everyday-wear glasses with cameras, mics, AI assistants, and sometimes a small heads-up display — worn all day in public, unlike the immersive headsets in Virtual Reality Design/Augmented Reality Design. Defining tensions: near-zero visual bandwidth, voice as primary channel, bystanders who never consented. Almost no peer-reviewed UX research exists yet (Glass-era social-acceptability work aside); this file is verified product behavior plus emerging convention.

Device classes (mid-2026 landscape — verify before citing specs)

Camera + audio, no display: Ray-Ban Meta Gen 1 (2023) / Gen 2 (Sept 2025 — longer battery, 3K video, "Conversation Focus" speech enhancement); Oakley Meta HSTN and Vanguard (2025, sport). The volume leaders: voice ("Hey Meta"), capture button, temple-stem touch, open-ear speakers, companion phone app. Amazon Echo Frames remain audio-only; Amazon camera glasses are reported for delivery drivers (~mid-2026), consumer later — announced, not shipped.
Monocular display + neural input: Meta Ray-Ban Display (Sept 2025, $799) — 600×600 px, ~20° field of view, projected into the lower-right of the right lens, up to 5,000 nits (sunlight-legible), ~6 h battery; ships with the Meta Neural Band sEMG wristband. The current reference design for display glasses.
Lightweight binocular HUD: Even Realities G1 (45 g, monochrome green text HUD, ~~$600) and Halliday (~~$500) — "normal-looking" glasses that display only sparse text (notifications, teleprompter, translation).
Android XR: Samsung Galaxy XR (Oct 2025) shipped, but it is a headset — see Virtual Reality Design/Augmented Reality Design. The Google/Samsung glasses (Warby Parker, Gentle Monster frames) shown at I/O 2026: Gemini audio glasses slated fall 2026, display glasses 2027 — as of July 2026, announced, unshipped.
Tethered virtual-display glasses (Xreal One/One Pro, Viture) are a different category — external-monitor optics for seated media/work.

Interaction model

Voice-first, AI-native: wearable AI assistants — wake word plus multimodal query ("Hey Meta, what am I looking at?") is the primary interface. Voice & Conversational Interfaces applies in full (Grice's quantity maxim doubly: answers play aloud in public); Designing AI-Powered Interfaces covers the probabilistic output — the assistant describes what the camera sees confidently, is sometimes wrong, and display-less models offer no screen to verify.
Temple-stem touch: tap/hold/swipe on the glasses arm — volume, play/pause, assistant. Small vocabulary, zero visual affordance, discoverable only via onboarding (emerging convention: ≤4 gestures).
EMG micro-gestures (Neural Band) — Meta's shipped set: thumb-index pinch = select; thumb-middle pinch = home/back (double = display toggle); thumb swipe along the index finger = d-pad scroll; pinch-and- rotate-wrist = volume/zoom dial. Works with the hand at your side or in a pocket — no raised arm, no camera view, socially invisible; sidesteps gorilla arm (Virtual Reality Design) and the social cost of public gesturing. CES 2026 added sEMG handwriting (trace letters on your leg) — still maturing.
Fallback chain: anything longer than seconds hands off to the phone app (Wearables & Smartwatch Design's keyboard rule, stricter — no keyboard exists). On display-less models the only feedback is earcons, the assistant's voice, and a status LED — every state must be distinguishable ear-only.

Glanceability at the extreme (display models)

The Ray-Ban Display puts content off-center (lower-right of one eye) — deliberately not in the line of sight; the display is off by default, summoned for seconds. One card at a time, no persistent HUD: this is Wearables & Smartwatch Design's two-second rule with less real estate — one string, one card, one action.
Monocular displays add binocular rivalry and focus-switching cost — users must glance, never "read." Outdoor legibility drives hardware (5,000 nits) and design: high contrast, large text, no dense layouts. Meta has published no formal display-glasses design spec; these rules are inferred from the shipped OS and dev docs — emerging conventions.

Session & attention design

World-first: the wearer is walking, driving, cooking, mid- conversation. Never demand sustained attention; every interaction must be interruptible, resumable, and completable hands- and eyes-free — Augmented Reality Design's safety principle at higher stakes, since the device is worn all day, not in a chosen AR session.
Notification triage stricter than the watch (Wearables & Smartwatch Design): a watch buzzes a wrist; glasses speak into your ear or light your vision. Emerging convention: default to digest/silence, surface only chosen people, never interrupt detected conversation.
Design around capture and query moments: photo/video, "what am I looking at" multimodal queries, live translation/captions, navigation glances, message triage — each a seconds-long episode, not a session.

Google Glass (2013–15) is the canonical cautionary tale: the "glasshole" backlash — bar bans, confrontations — was driven less by the display than by an always-ready camera with unclear recording state. Social acceptability is a launch requirement, not polish.
Capture LED norms: Ray-Ban Meta glasses light an outward LED during capture and refuse to record when it is covered (tamper detection); Gen 2 enlarged the LED. Reality check: $60 mods defeat it, and US bills to mandate recording indicators have been proposed — treat the LED as etiquette signaling, not enforcement.
Recording etiquette (emerging convention): verbal consent when pointing capture at individuals; none in bathrooms/gyms/medical spaces; app design that discourages stealth (audible shutter, visible gestures).
Private audio isn't: open-ear speakers leak — assume a neighbor can hear; keep spoken responses terse and non-sensitive (never read 2FA codes or message bodies aloud by default). The Display's panel has minimal outward light leak so captions stay private — preserve that.
Wake words in public carry social cost; offer non-voice invocation (button, stem tap, EMG pinch) for every voice-reachable action.

Accessibility: glasses as assistive tech

Be My Eyes on Ray-Ban Meta / Oakley Meta (partnered Sept 2024, live Nov 2024; expanded March 2026 to trusted-contact and company-support calls): blind and low-vision users get hands-free video assistance via camera and open-ear audio; Aira offers professional visual interpreters on the same hardware. Hands stay free for cane or guide dog — a real advantage over phone-based assistance.
Live captions on Ray-Ban Display: real-time in-lens transcription, mic-focused on the person you're facing, eye contact preserved; Gen 2's Conversation Focus amplifies in-person speech for hearing support.
Neural Band research (Meta + Univ. of Utah, 2026) targets limited hand mobility — sEMG needs faint muscle signals, not full movement. See Inclusive Design: mainstream-as-assistive, live.

Design guidance sources (what actually exists)

Meta Wearables Device Access Toolkit (developer preview Sept 2025; wearables.developer.meta.com) — camera/audio access from companion apps; display-glasses building opened late 2025; publishing limited to select partners during preview. An SDK with best-practice notes, not a design language — no Meta HIG-equivalent exists yet for glasses.
Android XR design guidelines (developer.android.com/design/ui/xr) — real and current, but written for headsets and wired XR glasses; design guidance for everyday audio/display glasses was still emerging pre- launch as of July 2026. Beyond these, practice is set by shipped products and community convention — expect it to change.

Sources

Meta Newsroom (Sept 2025) — "Meta Ray-Ban Display: AI Glasses With an EMG Wristband" (about.fb.com); Meta Quest Blog (Jan 2026) — CES 2026 Display update; Meta AI Glasses Help (meta.com/help) — Neural Band gestures, live captions, privacy/capture LED.
Meta for Developers — Wearables Device Access Toolkit (developers.meta.com/wearables; wearables.developer.meta.com).
Google — Android XR design guidelines (developer.android.com/design/ui/ xr); Google blog (May 2026, I/O) — Gemini eyewear timeline.
Be My Eyes newsroom (2024–26) — Meta partnership and expansion (bemyeyes.com/news); Aira — Meta glasses integration.
Shipped-behavior claims (2025–26): UploadVR, Engadget, Gizmodo (Display, Neural Band, LED mods, recording-indicator bills); Tom's Guide (Even Realities G1); UploadVR landscape ("Everyone Is Making Smart Glasses").
Google Glass reception: contemporaneous coverage 2013–15; "glasshole" framing is documented history, not formal research.