Subprocessors

Last updated: May 12, 2026

We use the following third-party service providers (sub-processors) to help deliver The Digital Clinic. Each processes data on our behalf under appropriate data protection agreements.

Infrastructure & Services

These sub-processors are used for all customers:

AI Providers

The Digital Clinic uses AI for features like clinical note generation, transcription, translation, knowledge-base search, and optical character recognition (OCR). Some providers are customer-configurable; others are platform-level decisions baked into the architecture. We keep them separate here so you know exactly what's selectable and what isn't.

Customer-selectable LLM providers

For chat, agent execution, clinical note generation, information extraction, knowledge classification, drug matching, language detection, and message translation, you choose which provider(s) we use for your organization. Your configured allowlist is enforced in code before any LLM request leaves the app — a misconfiguration fails loudly and is logged, not silently bypassed.

Customer-selectable OCR provider

For OCR on uploaded documents, we use Mistral OCR (EU/France). When your allowlist excludes Mistral, OCR is skipped and we fall back to local text extraction (Apache Tika) inside our own application — no document bytes are sent to any external provider.

Platform-level providers (not customer-selectable)

Two features rely on a specific provider for architectural reasons (audio pipeline design, shared embedding schema). They cannot be swapped per-customer; if a particular provider is incompatible with your requirements, the corresponding feature can be disabled for your organization at onboarding instead.

  • Audio transcription — Gladia (EU, France) — DPA. Used to convert recorded audio (in-app dictation, VoiceIt satellite) into text. Audio is sent to Gladia, transcribed, and the result is returned to us via webhook. Gladia retains audio only for the duration of the transcription job and does not train models on customer audio.
  • Knowledge-base search — OpenAI (embeddings only) — US — DPA. Converts text into vectors for similarity search. The vector dimensions are fixed in the database schema, so supporting alternative embedding providers would require a platform migration rather than a configuration toggle.

EU-only processing

If you require EU-only processing, restrict your customer-selectable allowlist to Mistral AI (France). Combined with our platform-level Gladia transcription (also France), this keeps all customer-content AI processing within the EU — except knowledge-base search, which uses OpenAI embeddings in the US. Knowledge-base features can be disabled for your organization on request.

Changes to This List

We keep this list up to date on our website. If you have concerns about a sub-processor, please contact us at team@digitalclinic.vet.

Contact

Clean Wake AB • Sprängarvägen 27 • 184 70 Åkersberga • Sweden • Email: team@digitalclinic.vet

Back to home