Voice Mobile App Privacy Policy

1. Introduction

This privacy policy applies to the Outermind voice mobile application (the "Voice App") distributed through Apple TestFlight and the Google Play Store internal track. It supplements the Outermind Privacy Policy and describes data practices specific to the mobile voice experience.

Outermind Inc. ("Outermind," "we," "us," or "our") is the data controller for personal data processed through the Voice App. Where this policy and the main Outermind Privacy Policy conflict, this policy governs Voice App data; where this policy is silent, the main policy applies.

2. Information We Collect

The Voice App collects the following categories of information:

Audio (microphone): Real-time microphone input is captured while a voice session is active. Audio is streamed for processing and is not stored on our servers as a recording. See section 4 for the full audio data flow.
Account identifier: When you sign in, your Outermind account ID and authentication token are used to link the Voice App session to your existing Outermind account.
Session metadata: Start and end timestamps, session duration, audio round-trip latency, and connection quality metrics. Used for the beta latency benchmark and service reliability.
Voice intents: The transcribed and structured intents derived from your speech (for example, "send an email to ...") are forwarded to the Outermind PA backend for execution and are subject to the main Outermind Privacy Policy from that point forward.
Device and diagnostic data: Device model, OS version, app version, and crash diagnostics. Used to diagnose bugs and improve stability during the beta.

3. Permissions We Request

The Voice App requests only the permissions required to deliver the voice experience:

Microphone: Required. Used solely to capture your speech during an active voice session. The microphone is not accessed when no session is active.
Network: Required. Used to stream audio to our infrastructure and receive synthesized voice responses.
Notifications (optional): Used to alert you when a background task completes or when your authentication needs to be refreshed. You may decline this permission with no impact on core voice functionality.

The Voice App does not request access to your contacts, photos, location, calendar, or any other on-device data.

4. How Your Audio Is Processed

When you start a voice session, your audio takes the following path:

Capture on your device: Audio is captured by the Voice App using the operating system's standard microphone APIs.
Real-time transport: Audio is streamed over an encrypted WebRTC connection to a self-hosted LiveKit server operated by Outermind in the Microsoft Azure East US 2 region.
AI inference: The LiveKit server forwards the audio stream to OpenAI's Realtime API (GPT-4o speech-to-speech model) for transcription and response generation.
Response playback: Synthesized voice audio is returned over the same encrypted connection and played back through your device.
Intent dispatch: When your speech contains an actionable intent (for example, an email or calendar request), the structured intent is forwarded to the Outermind PA backend over an authenticated HTTPS connection for execution.

We do not store recordings of your voice sessions. We do not use your audio to train machine learning models. See section 5 for the agreements that govern third-party processing.

5. Third-Party Services

The Voice App relies on the following third-party services to deliver real-time voice:

Microsoft Azure (East US 2): Hosts the self-managed LiveKit server that handles WebRTC audio transport. Subject to Microsoft's data processing agreement and our regional data residency commitments.
LiveKit (self-hosted): Open-source WebRTC media server, deployed and operated by Outermind. LiveKit Inc. does not have access to your audio.
OpenAI Realtime API (GPT-4o): Provides speech-to-speech inference. During the beta phase, we operate under OpenAI's zero-data-retention configuration for Realtime API traffic, meaning OpenAI does not retain your audio or transcripts after processing and does not use them to train models.
Apple TestFlight / Google Play (internal track): Distribute the beta build and collect crash diagnostics. Subject to the privacy policies of Apple Inc. and Google LLC respectively.

A complete and current list of Outermind sub-processors is available at /legal/sub-processors.

6. Data Retention

Data collected through the Voice App is retained as follows:

Raw audio: Not retained. Streamed for processing and discarded after the session ends.
Transcripts: Not retained by OpenAI under the zero-retention configuration. Outermind retains transcripts only as part of the audit log of executed intents, subject to your account's audit log retention tier (see the main Privacy Policy, section 8).
Session metadata and latency metrics: Retained for up to 90 days during the beta to support performance benchmarking and service reliability, then aggregated and de-identified.
Crash diagnostics: Retained for up to 180 days, then deleted.
Account termination: When you terminate your Outermind account, all Voice App data is deleted within 30 days, in line with the main Privacy Policy.

7. Data Security

The Voice App protects your data through:

End-to-end TLS 1.3 transport for control traffic and DTLS-SRTP encryption for WebRTC media
JWT-based authentication that links each session to a verified Outermind account
Short-lived access tokens issued per session with automatic expiration
Server-side encryption at rest (AES-256) for any persisted metadata
Tenant-level isolation that prevents one organization's session data from being accessible to another

8. Beta Status and Changes

The Voice App is currently in internal beta. Features, latency targets, and infrastructure may change as we iterate. Material changes to data handling will be communicated via:

An updated "Last updated" date at the top of this page
An in-app notification on the next session start, where the change affects beta participants
Email to the address associated with your Outermind account, for changes that materially expand the categories of data collected

9. Children's Privacy

The Voice App is not directed to children under 16. We do not knowingly collect personal data from children under 16. If you believe a child has used the Voice App, please contact us at privacy@outermind.ai so we can delete the data.

10. Your Rights

The rights described in section 13 of the main Outermind Privacy Policy apply equally to data processed by the Voice App, including the right to access, correct, delete, or export your data, and the right to lodge a complaint with a supervisory authority.

To exercise these rights, contact privacy@outermind.ai.

11. Contact

Questions about this Voice App Privacy Policy can be directed to privacy@outermind.ai. Security concerns should be reported to security@outermind.ai.