Subprocessors
Last updated: 2026-10-20
The following vendors help AudioMap process your data. By default, those handling audio and transcripts run on European infrastructure; the exceptions (payments, notifications, website assistant, browser speech recognition) are stated in each row.
Vendor due diligence per AEPD guidance (April 2026)
On 20 April 2026 the Spanish DPA (AEPD) published specific guidance on AI voice transcription requiring documented vendor due diligence: actual jurisdiction, subprocessors, retention policy, and model training policy. This page meets that requirement. Full mapping at AEPD compliance.
| Provider | Purpose | Location | DPA |
|---|---|---|---|
| Cloudflare R2 | Storage of audio files uploaded by users. | European Union (explicit EU jurisdiction). | View DPA |
| AssemblyAI | Automatic audio transcription. | European Union (endpoint api.eu.assemblyai.com). | View DPA |
| Local Whisper (faster-whisper) | Self-hosted fallback transcription on our own servers. No outbound traffic. | Germany (Hetzner DE server under AudioMap's direct control). | n/a: self-hosted processing, not an external subprocessor |
| Google Cloud (Vertex AI) | Language models (Gemini) for AI Notes, chat, analysis and generation of deliverables (emails, documents, slides). | European Union (Vertex AI EU locations). | View DPA |
| Mistral AI | Backup language model when the main provider is unavailable: chat, AI Notes, deliverables, marker topics, chapter index and follow-up suggestions. It only comes into play if the main provider fails. It is not used to train models. | France (European Union). | View DPA |
| Cerebras Systems Inc. (via AgenteUno) | Language model of the website's conversational assistant (AgenteUno). Processes chat messages as a processor. It does not process audio or transcriptions of your notes. | US. Not listed under the EU-US Data Privacy Framework (DPF): the transfer relies on the standard contractual clauses (Commission Implementing Decision (EU) 2021/914) incorporated into its DPA. | View DPA |
| Hetzner Online GmbH | Dedicated servers (Postgres, Redis, workers, local Whisper). | Germany (Falkenstein / Nuremberg). | View DPA |
| Sentry | Error tracking and observability. | Region not pinned by AudioMap: it depends on the service configuration. | View DPA |
| PostHog | Product analytics, heatmaps and masked session recording. Only if configured and with consent in the cookie banner. | Region not pinned by AudioMap: it depends on the service configuration. | View DPA |
| Browser speech recognition (Google in Chrome/Edge, Apple in Safari) | Only in the Live Assist panel, and only while you have it on: live recognition is run by the browser's own service, not by AudioMap. The transcription of your notes does NOT go through here. | Outside the EU (US), under the browser vendor's policy. AudioMap is not a party to that transfer and does not receive that audio. | n/a: your browser sends it; documented for transparency |
| Google Firebase Cloud Messaging and Apple Push Notification service | Delivery of mobile app notifications. They receive the device identifier and the text of the alert (for example, the title of the note that is ready); never the audio or the transcription. | US (Google LLC and Apple Inc.). | View DPA |
| Stripe | Payment processing. Does NOT process audio content. | US (payment data only: no audio, no transcriptions). | View DPA |
We keep this page up to date. If we add or change a subprocessor, we will notify B2B customers with a signed DPA at least 30 days in advance.