[ LEGAL ]Subprocessors

Subprocessors

Updated 22 Aug 2026 · Vaidra

Vaidra uses the following vendors to operate the marketing site and the Platform. Patient encounter content (including audio, transcripts, and clinical text) is processed by speech and language vendors when those features run. This list may change; we will update this page.

01

Infrastructure and product

  • Amazon Web Services

    API hosting, networking, and PostgreSQL (clinical/FHIR data) with encryption at rest enabled on the database.

    Region: AWS regions we deploy to (may be outside India)

  • Vercel

    Hosts the marketing site and clinician application.

    Region: Global edge

  • Cloudflare R2

    Object storage for consult audio, encounter attachments, and public assets such as a product walkthrough video.

    Region: Cloudflare storage locations

  • Qdrant

    Vector index for medicines, terminology, and institutional blueprints (catalogue and template text — not the live patient chart).

    Region: Our private deployment / configured host

  • Axiom

    Operational logs, tracing, and optional browser fatal-error reports. Reports may include route, an encounter identifier, and error text.

    Region: Axiom cloud

  • Google

    Clinician sign-in (OAuth). Google Speech-to-Text when that engine is selected. Optional Gemini / Vertex models if enabled for a workspace.

    Region: Google Cloud

02

Speech and language

These providers receive consult audio and/or clinical text when the corresponding feature is on. Optional engines run only if selected or configured. They process data under their own terms; we do not claim they never use data to improve their services.

  • OpenAI

    Language models for drafts and chat; embeddings for medicines, terminology, and blueprints (not live encounter notes); Whisper speech-to-text when that engine is selected.

    Region: United States / OpenAI processing locations

  • Sarvam AI

    Streaming speech-to-text (including Indian languages) when that engine is selected.

    Region: Sarvam processing locations

  • Deepgram

    Streaming speech-to-text (including a medical English mode). In some deployments the browser may connect to Deepgram directly.

    Region: Deepgram processing locations

  • Anthropic

    Optional

    Language models for drafts and chat.

    Region: Anthropic processing locations

  • Microsoft Azure OpenAI

    Optional

    Language models when an Azure deployment is configured.

    Region: Chosen Azure region

  • NVIDIA

    Optional

    Language models if enabled in configuration.

    Region: NVIDIA cloud

  • OpenRouter

    Optional

    Routed access to language models if enabled in configuration.

    Region: OpenRouter / upstream model regions

  • ElevenLabs

    Optional

    Voice features if enabled in configuration.

    Region: ElevenLabs cloud

03

Marketing and communications

  • Calendly

    Demo scheduling when you follow a link to book a walkthrough on Calendly’s site.

    Region: Calendly cloud

  • Google Analytics 4

    Marketing-site traffic measurement when a measurement ID is set.

    Region: Google

  • Microsoft Clarity

    Session analytics on the marketing site and, if configured, the Platform.

    Region: Microsoft

  • Email / SMTP provider

    Optional

    Org invites and transactional email when email is enabled.

    Region: Configured mail host

04

Notes

We do not currently use a card payment processor. Paid access is arranged with us. For how these vendors fit DPDP roles, see the Privacy Policy. Security overview: Security.

Questions or a vendor change notice request: hello@vaidra.care.