Meta Wearables
Vanguard · Gen 1 · Gen 2 · 3K Ultra HD · bi-directional audio
Three launch surfaces
One intelligence layer.
Different ways into the world.
SecondSee works like a spatial studio: begin with the edge, shape the system around the launch, then deliver it through the right hardware and interface.
01
Commercial systems
Put perception inside the operation.
Launch real-time viewing, tracking, recording, and operator workflows around the work already happening at the edge.- Live operations
- Workflow orchestration
- Multi-publisher rooms
02
Developer platforms
Compose intelligence instead of rebuilding it.
Start with device clients, a binary relay protocol, and a visual workflow engine spanning local vision, audio, models, and outputs.- Composable DAGs
- Device SDK surfaces
- Developer tools
03
Consumer experiences
Make the interface disappear into the moment.
Build camera, microphone, speaker, and display experiences around wearable form factors people already understand how to use.- Natural capture
- Audio interaction
- Wearable display
The system
A live loop from the world
to intelligence and back.
Bring your own compute hub. SecondSee connects capture, perception, orchestration, and output without forcing the whole experience into one device.
01
Capture
See and hear
Glasses and device cameras become live camera, microphone, and speaker peripherals.
H.264 · PCM · sensors02
Perceive
Process at the edge
Enhance, detect, track, classify, transcribe, and record through parallel device-side stages.
Vision · Core ML · OC-SORT03
Orchestrate
Compose the workflow
A relay and DAG engine route live signals through models, rooms, triggers, and services.
FRLY · FRAU · FRSE04
Deliver
Return intelligence
Send the result to operator dashboards, recordings, audio, overlays, and wearable displays.
Web · device · displayINSIDE THE STUDIO
Watch the system think.
A live camera signal moving through detection, tracking, OCR, and an operator-facing viewer.
Live Camera Relay
Stream from supported wearable publishers or a desktop camera through the shared relay and into the web viewer.
Detection & Tracking
YOLO detection, OC-SORT tracking, OCR — chain nodes into a live pipeline that runs while you stream.
Workflow Pipeline
Compose detection, tracking, models, triggers, recordings, and outputs as a live graph shared by device and web surfaces.
Studio capabilities
A working system,
not a concept reel.
The live remote call is the headline. Every surface below is in the tree today — capture, relay, workflows, operator tools — and it all exists to keep that call standing up in the field.
Wearable capture
Stream camera and audio from Ray-Ban Meta on iOS, INMO on Android, or a desktop camera into the same relay — the live feed your expert watches from a browser.
On-device perception
Enhance frames, run Vision and Core ML, detect objects, track identities, classify scenes, and record locally.
Visual workflows
Compose sources, processors, triggers, models, and outputs as a live DAG instead of hard-coding one assistant.
Live expert calls
One call, two ends — the worker's glasses and the expert's browser — carried by a shared binary protocol with multi-publisher room support.
Speech + audio
Route phone and glasses microphones, transcription, audio classification, synthesized speech, and playback.
Operator platform
Monitor live streams, edit workflows, inspect telemetry, manage recordings, and work with developer tools on the web.
Device field
Build for the surface
that fits the launch.
SecondSee separates the intelligence layer from the frame on someone's face. Shipping integrations stay distinct from development and evaluation work.
Available now
Shipping
INMO Air 3S
Android publisher · camera and audio modules · shared relay protocol
Display & spatial
Meta Ray-Ban Display
Native display capability and web display surfaces built; hardware validation continues
Apple Vision Pro
visionOS publisher target and spatial viewing surfaces
Samsung Galaxy XR
Camera and display capability probes; not presented as a production surface
Under evaluation
Additional wearables
Lensmoo, HeyCyan, and Mentra remain evaluation surfaces rather than compatibility promises
Straight answers
What works today.
What doesn’t yet.
The headline is the live remote call — glasses in the field, an expert in a browser, one shared view of the job. The answers below cover that, and keep the rest honest: shipped means shipped.
It's a live remote call for camera glasses. The worker streams what they see; an expert watches from a browser and talks them through the job. The same feed powers recognition, guidance, transcription, and local recording.
Yes — that's the core call, and it's live today. The worker's glasses publish first-person video with two-way audio to the relay; the expert joins from any browser, no app to install. What the expert sees is what the worker sees, in real time.
iOS with Ray-Ban Meta and Android with INMO are the furthest along. macOS, visionOS, browser, and XR builds are in the tree at earlier stages. The device field above keeps the two apart — if it isn't shipping yet, it doesn't say it is.
No. Capture, codecs, audio, sensors, and outputs vary by glasses model and hardware. A workflow that runs on one device may need adjusting on another — check the device's capabilities before you plan around it.
Wherever the job needs it: on the streaming device, a local compute hub, the relay, or an external model service you configure. You choose the split per workflow — it doesn't all have to live on one device.
Deployments run with authenticated sessions, access controls, share tokens, gateway isolation, TLS, and your own object storage settings. Retention, encryption, and compliance specifics depend on how you deploy it — we'd rather walk through your setup than promise defaults we don't control.
Teams whose people do hands-on work — the factory floor, inspections, field service — and need an expert or AI in the loop now. Start with one device, one signal, one outcome you can name, not a company-wide rollout.
Start with a live call
Put your expert
in the field.
Bring a device and a workflow. We'll start with a live remote call — your expert watching what the worker sees — then map where perception and automation take over.
Start a conversation
Include the target hardware and the live signal you need to use.