Sub-processors

Last updated: 16 July 2026

We use the following sub-processors to deliver the AI Fortis Voice AI Platform. We update this list when it changes and give clients prior notice of additions or replacements in accordance with the DPA.

• LiveKit — real-time voice infrastructure (call media, WebRTC/SIP).
• Google — AI language models (Gemini) powering agent conversations; Google Workspace for service email and calendar scheduling.
• ElevenLabs — speech synthesis (text-to-speech).
• Deepgram — speech recognition (speech-to-text).
• OpenAI — speech recognition and, on agents configured for it, real-time speech-to-speech conversation (receives call audio).
• Microsoft (Azure Speech) — speech recognition on agents configured for it (receives call audio).
• Twilio, Telnyx and other telephony carriers — phone numbers and call carriage (PSTN/SIP).
• Supabase — application database and authentication (hosted in the EU).
• Vercel — application hosting.
• Cloudflare (R2) — call recording storage.
• Upstash — queues, rate limiting and concurrency control.
• Resend — transactional email (if enabled).
• Anthropic — the built-in assistant used by our own staff. It receives platform metadata (agent, campaign and client names and operational statistics). It does not receive call recordings, call transcripts or end-customer phone numbers.
• Sentry — application error monitoring. Diagnostic data only; identifying fields are removed before transmission.

Which speech providers apply to a given agent depends on how that agent is configured; your agent's configuration is available from us on request.

Each sub-processor is engaged under a written agreement requiring appropriate security and GDPR compliance. Where a provider processes personal data outside the EEA, the EU Standard Contractual Clauses and, where needed, supplementary measures apply. Per-vendor processing locations and transfer details are available on request from office@aifortis.com.