Back to Home
Compliance Center

Privacy Policy

Last Updated: June 15, 2026

1. Overview

Valyrian Voice is committed to protecting the privacy of your organization, your users, and the individuals called or calling through our telephony voice systems. This Privacy Policy details how we handle, process, and secure voice streams, transcripts, call logs, and customer metadata within our dedicated voice AI services.

2. Data Security & Storage Boundaries

We prioritize data isolation and compliance. Unlike traditional multi-tenant API integrations, Valyrian Voice processes conversational workflows inside partitioned environment runtimes:

  • Single-Tenant Storage: Client campaign databases, phone contact lists, and API keys are isolated in single-tenant data structures, eliminating data leakage concerns.
  • Stateless GPU Inference: Conversational models (ASR, Reasoning, TTS) run in-memory on dedicated GPU instances and do not persist client data or call contents.
  • Optional Log Storage: Customers have full control over the lifecycle of call logs, audio recordings, and transcripts, which can be deleted automatically post-sync to CRMs.

3. Information We Collect

To provide our voice automation services, we process the following types of information:

  • Account Information: Organization name, admin credentials, billing details, and API configuration credentials.
  • Telephony Metadata: SIP trunk route addresses, call duration metrics, phone numbers, and dial status codes.
  • Voice Processing Data: Duplex PCM audio streams processed dynamically for transcription, context reasoning, and real-time audio synthesis.

4. GDPR and Regulatory Compliance

For users operating in European Jurisdictions, Valyrian Voice processes call data as a Data Processor under instructions from our clients (the Data Controllers). We support:

  • Automatic data redaction rules for transcripts.
  • Enforcing local Do-Not-Call (DNC) list verification prior to outbound call generation.
  • Regulatory hour-of-day calling restrictions built into our Dial Session engines.

5. Sub-processors

When utilizing GPU execution nodes, all calculations are performed locally on host instances. For clients choosing the BYO API model, audio and text data are transmitted to selected external providers (such as Deepgram, OpenAI, or ElevenLabs) in accordance with the keys configured by the client.

6. Contact Information

For compliance questions, data access requests, or deletion protocols, contact us at contact@valyrianvoice.com.