Subprocessors

Last updated: September 24, 2026

Not A Number Labs Incorporated ("Chorus," "we," "us," "our") works with a small number of trusted third parties, called sub-processors, that help us operate the Service. A sub-processor is a company we use to process customer data on our behalf, for example to host our systems, store data, or generate the agent's responses. This page lists the sub-processors we currently use and what each one does.

1. Current sub-processors

The companies below process customer data on our behalf to deliver the Service. Unless noted otherwise, they process data in the United States.

Infrastructure and platform

  • Amazon Web Services: Cloud hosting and infrastructure for our application.
  • Vultr: Hosts the dedicated virtual machine each agent runs on, including the agent's workspace, memory, and files.
  • DigitalOcean: Hosts the service that routes and meters the agent's requests to model and capability providers.
  • Neon: Managed PostgreSQL database where customer data is stored.
  • Render: Hosts the service that routes messages between platforms such as Slack and the agent.
  • Google Cloud: Stores the software images our agents run on and delivers Gmail change notifications for connected accounts.
  • Vercel: Hosts our website.
  • Cloudflare: Stores files for the iOS app building feature.
  • Clerk: Authentication and user account management.
  • Nango: Manages the OAuth connections between an agent and the third-party tools a customer chooses to connect.

Messaging

These providers carry messages between the agent and the people it talks to, depending on the channels a customer connects.

  • Linq: Sends and receives iMessage, SMS, and RCS messages.
  • Kapso: Sends and receives WhatsApp messages.
  • Slack: Sends and receives messages in connected Slack workspaces.
  • Microsoft: Sends and receives messages in connected Microsoft Teams workspaces.

Billing, analytics, and communications

  • Stripe: Processes payments.
  • Autumn: Manages plans, usage credits, and billing.
  • PostHog: Product analytics, session recordings of our website, and application logs.
  • Datadog: Application monitoring and logs.
  • Resend: Sends transactional emails such as billing notices.

Model providers

These providers process the message content we send to the model in order to generate the agent's response. Which one is used depends on the model a customer selects for their agent.

  • Anthropic: Provides large language models that power the agent. Anthropic does not train its models on this data.
  • OpenAI: Provides large language models that power the agent.
  • Google: Provides the Gemini large language models that power the agent.
  • xAI: Provides the Grok large language models that power the agent.
  • Groq: Provides accelerated model inference for supported models.
  • Baseten: Provides hosted model inference for supported models.
  • OpenRouter: Routes model requests to supported model providers on our behalf.
  • Moonshot AI: Provides the Kimi large language models for supported models. May process data outside the United States.
  • DeepSeek: Provides the DeepSeek large language models for supported models. May process data outside the United States.

Capability providers

These providers receive data only when a customer or the agent uses the specific capability they provide.

  • ElevenLabs: Converts text to speech for voice features.
  • SerpApi: Returns web search results in response to an agent's search query.
  • Firecrawl: Fetches and extracts the content of web pages an agent visits.
  • Browserbase: Provides hosted headless browser sessions for agent browsing tasks.
  • Deepgram: Transcribes voice messages sent to an agent.
  • Exa: Returns web search results in response to an agent's search query.
  • Supadata: Extracts transcripts and content from videos and web pages an agent is asked to read.
  • ScrapeCreators: Returns public social media profiles and posts an agent looks up.
  • Foreplay: Returns public ad library data an agent looks up.
  • AppKittie: Returns public app store data an agent looks up.
  • OpenWeather: Returns weather data for locations an agent looks up.
  • Alpha Vantage: Returns financial market data for symbols an agent looks up.
  • fal and Ideogram: Generate images when an agent is asked to create one.
  • Microsoft Azure, Limrun, and Apple: Build, preview, and publish iOS apps when a customer uses the iOS app building feature.

2. How we choose sub-processors

Before a sub-processor handles customer data, we review how it stores and protects that data and confirm it offers terms consistent with the commitments in our privacy policy. We give each sub-processor only the data it needs to perform its function.

3. Changes to this list

We keep this page up to date when we add or remove a sub-processor that handles customer data. We encourage you to check back here for the current list.

4. Contact

If you have questions about our sub-processors, email us at support@chorus.com.