EU sovereignty

A sovereign platform,
your choice of models

Our orchestrator, dashboard, and API run on EU infrastructure, with platform data stored in the EU. Our infrastructure providers hold ISO 27001 and/or SOC 2 Type II certifications; the per-provider overview is in the Trust Center. The AI models you plug into it are your call. We help you pick the most compliant stack for your use case.

What's actually sovereign

The platform layer, fully European

The platform layer runs in EU regions: orchestration, storage, and control-plane traffic are processed and stored inside the European Union. Our infrastructure providers hold ISO 27001 and/or SOC 2 Type II certifications; the per-provider overview is in the Trust Center.

Orchestrator on EU compute

Voice orchestration, session state, and call routing run on EU-resident infrastructure end-to-end.

Recordings & transcripts in EU

All persistent data — call recordings, transcripts, analytics — stored in EU regions only.

Dashboard & API in EU

The management plane, dashboard, and public API are served from EU regions.

Secrets in an EU vault

Customer keys and credentials are stored encrypted in a vault in the EU, isolated per organisation, and only used to run that organisation's calls.

Where we're transparent

The model layer is your decision

We integrate with the best providers in the world — many of them American. That's a trade-off you control. Pick whatever fits your latency, language, and compliance needs. We won't hide where the request is going.

US-based providers

OpenAI, whose GPT-Live is our preferred realtime model, plus Deepgram and other industry-standard providers. By default OpenAI processes GPT-Live audio outside the EU, and that is the trade-off. The processing location per model is listed in the Trust Center.

OpenAI GPT-Live · Deepgram · Inworld

European providers

Mistral, Gladia, ElevenLabs (UK / GDPR-aligned), or Gemini via Vertex AI on EU-West4 (Netherlands). The Trust Center lists where each one processes data.

Mistral · Gladia · Gemini EU-West4 · ElevenLabs UK

Models in your own environment

For zero external dependencies: local models running inside your own private cloud or on our on-premise unit, set up with you.

Private cloud · on-premise unit
Recommended full-EU stack

Need everything in Europe? Here's how

These combinations are built from providers that process in Europe; the Trust Center lists each provider's processing location and transfer basis. We advise you on the best fit for your language and use case.

FR

French / Western Europe

STT
Gladia
LLM
Mistral Large
TTS
ElevenLabs UK
NL

Dutch / multilingual

STT
Mistral Voxtral
LLM
Gemini via Vertex (EU-West4)
TTS
ElevenLabs UK
EU

Regulated / on-premise

STT
Local speech recognition
LLM
Local language model
TTS
Local voice

Combinations evolve quickly. Talk to us about what's currently best for your use case.

Technical safeguards

Security by default

Defaults that match what enterprise security teams want to see in their first review.

AES-256 encryption at rest

All persistent storage encrypted with industry-standard symmetric encryption.

TLS 1.3 in transit

Modern TLS for every API call, every webhook, every internal RPC.

HMAC-signed webhooks

Every webhook is signed so you can verify the request came from us — not a spoofer.

Tenant isolation

Per-organization data partitioning. Strict access controls and audit logging on every read.

Vault-managed secrets

Every credential lives in a dedicated secret store, never in environment dumps or logs.

BYOK for everything

Use your own API keys with model providers. They are stored encrypted and only used to run your calls; your providers bill you directly.

Want help picking the right stack?

Tell us your use case and we'll map the most compliant provider combination. The standard DPA is ready to sign at no additional cost; confidential security documents are available on request.