Subprocessors

Last updated August 25, 2026

echowin engages the third parties below to help deliver the echowin Service. Each one Processes Customer Personal Data (as defined in our Data Processing Addendum) on echowin's behalf and under written terms no less protective than those in the DPA.

To be notified in advance when we add or replace a subprocessor, subscribe using the form at the bottom of this page, or email privacy@echo.win and ask to be added to the subprocessor change notification list. We provide at least 30 days' notice before a new subprocessor begins Processing Customer Personal Data.

Infrastructure and hosting

SubprocessorLocation(s) of processingPurpose / description of processing
Cloudflare, Inc. (US)Global edge network; storage and databases run on Cloudflare's global/automatic region settingPrimary application platform. Runs the backend services and holds live call and chat session state; stores call recordings and audit logs; caches configuration and phone-number data; brokers all database connections; queues background jobs; collects application logs and performance telemetry; serves uploaded images; performs or proxies AI inference where a Cloudflare-hosted model is selected
Google LLC (Google Cloud Run, US)United StatesHosts the echowin web application, dashboard, and customer portal. Stores build artifacts and application secrets
Supabase, Inc. (US)United States. Supabase hosts on Amazon Web Services infrastructurePrimary production database (PostgreSQL). Stores call records, transcripts, call summaries and analysis, contact records, knowledge base content and vector embeddings, and account data

Telephony and messaging

SubprocessorLocation(s) of processingPurpose / description of processing
Telnyx LLC (US)United StatesCarries inbound and outbound voice calls, provides phone numbers, performs call control and call recording, and sends SMS/MMS messages. Processes caller and recipient phone numbers, call audio, call metadata, and message content

Speech-to-text (transcription)

SubprocessorLocation(s) of processingPurpose / description of processing
Deepgram, Inc. (US)United StatesReal-time transcription of live call and web-voice audio
OpenAI, L.L.C. (US)United StatesReal-time transcription of live voice audio via the OpenAI Realtime API
Google LLC (Google AI / Gemini API)United StatesReal-time speech-to-speech processing of live call audio, and post-call transcription, summarization, and quality analysis of call recordings. Receives the call recording audio file

AI model providers (conversation intelligence)

SubprocessorLocation(s) of processingPurpose / description of processing
OpenAI, L.L.C. (US)United StatesGenerates AI agent responses during calls and chats; generates embeddings for knowledge base content and search queries; generates speech audio
Anthropic PBC (US)United StatesGenerates AI agent responses during calls and chats; processes uploaded documents and knowledge base content; generates call insights
Google LLC (Google AI / Gemini API)United StatesGenerates AI agent responses; produces call summaries, transcripts, sentiment and quality analysis, and business-insight digests
Groq, Inc. (US)United StatesGenerates AI agent responses when a Groq-hosted model is selected for an agent
Microsoft Corporation (Azure OpenAI Service)United StatesGenerates AI agent responses when an Azure-hosted model is selected for an agent
OpenRouter, Inc. (US)United States, and the locations of the inference endpoints OpenRouter routes to. echowin restricts routing to endpoints that neither retain nor train on request content, but OpenRouter does not guarantee a fixed geography for themRoutes AI inference requests to third-party model endpoints. Used to extract knowledge base content from customer webpages and documents, summarize chat conversations, generate agent instructions and prompts during automated setup, and generate outbound-campaign content

Text-to-speech (agent voice)

Each of these Processes the text of the AI agent's spoken response, which may contain personal data (for example, a caller's name, address, or appointment details read back aloud).

SubprocessorLocation(s) of processingPurpose / description of processing
ElevenLabs Inc. (US)United StatesGenerates agent speech audio, including custom cloned voices
Microsoft Corporation (Azure AI Speech)United StatesGenerates agent speech audio
Google LLC (Cloud Text-to-Speech)United StatesGenerates agent speech audio
OpenAI, L.L.C. (US)United StatesGenerates agent speech audio
Deepgram, Inc. (US)United StatesGenerates agent speech audio
Speechify Inc. (US)United StatesGenerates agent speech audio
Theai, Inc. d/b/a Inworld AI (US)United StatesGenerates agent speech audio, including custom voices

Notifications, support, and analytics

SubprocessorLocation(s) of processingPurpose / description of processing
Resend, Inc. (US)United StatesDelivers transactional email, including call transcript and summary emails, notification emails, and account emails
Discord, Inc. (US)United StatesReceives internal operational alerts that include call metadata (partially masked caller and recipient phone numbers, call identifiers, agent and account names) and AI-generated call quality assessments
Mango Technologies, Inc. d/b/a ClickUp (US)United StatesStores support tickets and internal support documentation, which may contain customer-supplied details about a call or account
PostHog, Inc. (US)United StatesProduct analytics and error monitoring for the echowin dashboard and portal. Captures authenticated user identifiers, IP addresses, page and interaction events, and client-side exception payloads

Customer-procured services (not echowin subprocessors)

Where a customer connects their own account for one of the services below, that service is the customer's own vendor under Terms of Service §2.7. echowin transmits data to it at the customer's direction and under the customer's agreement with that provider.

Google Calendar · Microsoft Outlook / Microsoft Graph · Wix · Square · GoHighLevel · Shopify · Zapier · Slack · Discord (as a customer-connected channel) · Meta / WhatsApp Business · any customer-configured outbound webhook endpoint.

Subprocessor change notifications

We'll email you at least 30 days before a new subprocessor begins processing customer personal data.