Audio and camera glasses Coming soon

Glasses without a display: speech, spoken questions, voice and the camera

Models

Meta glassesPlanned

Meta's official device toolkit, starting with the Ray-Ban models it supports

Not supported yetAmazon Echo Frames · Solos

CapabilityMeta glasses
input.utteranceComing soon
capture.photoComing soon
capture.audioComing soon
present.speakComing soon
present.notifyComing soon
ask.confirmComing soon

Keys, connecting a user's device, permissions, routing, errors and test mode work the same for every category: see Projects, environments and keys, Connect a device, Permissions and safety, Errors and Test mode. The examples assume a dopl client and a deviceId from the Quickstart.

What your agent receives

Voice transcripts

input.utteranceT1

What the user says, as text.

Limits
  • Set by each device's descriptor. No model in this category is connected yet.
Permission
Scope input.*, ticked by default on the consent screen. At call time: allowed. See Permissions and safety.

Call

// No call returns input.utterance yet. It is planned as the `input.utterance` webhook.

Example response

// Not available yet.

Supported on

  • Meta glassesComing soon

Photos

capture.photoT2

A photo from the glasses' camera, taken when your agent asks. T2: the user must tick it on the consent screen.

Limits
  • Set by each device's descriptor. No model in this category is connected yet.
Permission
Scope capture.*, unticked on the consent screen until the user ticks it. At call time: allowed only when the user ticked it. See Permissions and safety.

Call

const { action_id } = await dopl.devices.action(deviceId, "capture.photo", {});
const action = await dopl.actions.get(action_id);
console.log(action.result);

Example response

{
  "action_id": "a6d1c0b2-7e45-4f19-8c3a-2b9e5d4f7a11",
  "device_id": "8b0c6f1e-3d2a-4c51-9a7e-2f4d6b1c9e30",
  "capability": "capture.photo",
  "status": "done",
  "result": {
    "simulated": true
  },
  "error": null,
  "created_at": "2026-10-27T17:05:00Z",
  "updated_at": "2026-10-27T17:05:09Z"
}

Supported on

  • Meta glassesComing soon

Audio clips

capture.audioT2

A recording from the glasses' microphones, taken when your agent asks.

Limits
  • Set by each device's descriptor. No model in this category is connected yet.
Permission
Scope capture.*, unticked on the consent screen until the user ticks it. At call time: allowed only when the user ticked it. See Permissions and safety.

Call

const { action_id } = await dopl.devices.action(deviceId, "capture.audio", {});
const action = await dopl.actions.get(action_id);
console.log(action.result);

Example response

{
  "action_id": "a6d1c0b2-7e45-4f19-8c3a-2b9e5d4f7a11",
  "device_id": "8b0c6f1e-3d2a-4c51-9a7e-2f4d6b1c9e30",
  "capability": "capture.audio",
  "status": "done",
  "result": {
    "simulated": true
  },
  "error": null,
  "created_at": "2026-10-27T17:05:00Z",
  "updated_at": "2026-10-27T17:05:09Z"
}

Supported on

  • Meta glassesComing soon

What your agent can send

Speak

present.speakT1

Text read aloud in the user's ear.

Limits
  • Set by each device's descriptor. No model in this category is connected yet.
Permission
Scope present.*, ticked by default on the consent screen. At call time: allowed. See Permissions and safety.

Call

const { action_id, status } = await dopl.devices.action(deviceId, "present.speak", {
  text: "Your 3pm moved to 3:30."
});

Example response

{
  "action_id": "a6d1c0b2-7e45-4f19-8c3a-2b9e5d4f7a11",
  "status": "done"
}

Supported on

  • Meta glassesComing soon

Notify

present.notifyT1

A short notification, spoken on glasses without a display.

Limits
  • Set by each device's descriptor. No model in this category is connected yet.
Permission
Scope present.*, ticked by default on the consent screen. At call time: allowed. See Permissions and safety.

Call

const { action_id, status } = await dopl.devices.action(deviceId, "present.notify", {
  title: "Build finished",
  body: "main passed"
});

Example response

{
  "action_id": "a6d1c0b2-7e45-4f19-8c3a-2b9e5d4f7a11",
  "status": "done"
}

Supported on

  • Meta glassesComing soon

Ask to confirm

ask.confirmT1

A yes or no question the user answers by voice. The answer comes back on the action as confirmed.

Limits
  • Set by each device's descriptor. No model in this category is connected yet.
Permission
Scope ask.*, ticked by default on the consent screen. At call time: allowed. See Permissions and safety.

Call

const { action_id, status } = await dopl.devices.action(deviceId, "ask.confirm", {
  question: "Send the reply to Ana?"
});

Example response

{
  "action_id": "a6d1c0b2-7e45-4f19-8c3a-2b9e5d4f7a11",
  "status": "done"
}

Supported on

  • Meta glassesComing soon