ATTENTION LABS

STANDBY AUDIO ONLY

Microphone

Camera

Speaker

Input level

ROOM —

floor: open

CAM

SELECTIVE AUDITORY ATTENTION

It knows when you’re talking to it, and when you’re not.

Allow camera and microphone.

Camera and mic let it tell when you're talking to it.

Reconnecting.

The connection dropped. Trying to pick it back up.

Choose.

Model

Voice

Instruction

OpenAI Key SAS Token *

Waking up

Initializing vision model
Initializing audio model
Fusing multimodal embeddings
Calibrating to your room
Warming up classifier

0%

NATIVE · OPENAI SERVER VAD

Silent

Confidence —

Answer threshold 85%

0.85

How sure it is that you're talking to it, measured live as you speak.

Selective Auditory Attention (SAA) is ON

Faces

0

Your voice

0%

Conversation

Idle

00000 Tokens saved

Try asking the AI: "Tell me about vacation ideas to Europe."

Silent To someone else To me

That’s the demo.

Run it again, or put the same model in your own stack.

Read the docs