Fieldwork AI Inc.
Sub-processors
The service providers Fieldwork uses to deliver its service.
Last updated: 12 September 2026
Fieldwork AI Inc. uses, or may use, the following sub-processors to provide the Service. We update this page before adding or replacing a sub-processor, and Users with the admin role in each customer Organization are notified of the change automatically through the Service. Capitalized terms have the meanings in our Terms of Service and Data Processing Addendum.
Sub-processors
| Sub-processor | Purpose | Data processed | Data at rest | Transfer mechanism |
|---|---|---|---|---|
| Amazon Web Services, Inc. | Cloud infrastructure (hosting, storage, databases, compute) and model inference via Amazon Bedrock | Customer Data, Account Data, Usage Data | USA (us-east-2) | EU SCCs + UK Addendum; EU–US Data Privacy Framework (EU–US DPF) |
| Google Cloud Platform (Google LLC) | Cloud infrastructure and model inference (Vertex AI) | Customer Data, Usage Data | USA | EU SCCs + UK Addendum; EU–US DPF |
| Microsoft Azure (Microsoft Corporation) | Cloud infrastructure, AI inference, and speech transcription | Customer Data, Usage Data | USA (East US) | EU SCCs + UK Addendum; EU–US DPF |
| Anthropic, PBC (Anthropic API Platform) | Large language model inference | Customer Data submitted as model context; Usage Data | USA (primary) | EU SCCs + UK Addendum; EU–US DPF |
| OpenAI OpCo, LLC (OpenAI Platform) | Large language model inference | Customer Data submitted as model context; Usage Data | USA (primary) | EU SCCs + UK Addendum; EU–US DPF |
| Fireworks.ai, Inc. (Fireworks AI) | Large language model inference on Fireworks infrastructure | Customer Data submitted as model context; Usage Data | No prompt or response content stored for our configured inference requests; usage metadata retained | EU SCCs + UK Addendum |
| Vercel, Inc. (Vercel AI Gateway) | Model request routing and failover between inference providers | Customer Data in transit as model context; Usage Data | None - transit only (usage metadata retained in USA; no prompt or response content stored) | EU SCCs + UK Addendum; EU–US DPF |
| OpenRouter, Inc. | Model request routing and failover between inference providers | Customer Data in transit as model context; Usage Data | USA and other processing locations (usage metadata; prompt and response logging off by default) | EU SCCs + UK Addendum |
| Composio Inc. | Hosted connectivity for third-party integrations: connection flow, credential storage and refresh, and tool-call proxying. Not used for Google connections, whose credentials Fieldwork stores directly | Credentials issued by the connected third-party system; Customer Data in transit as tool inputs and responses; pseudonymous user identifiers | USA (credentials and connection metadata; no conversation content stored) | EU SCCs + UK Addendum |
| Resend, Inc. | Transactional and notification email delivery (account emails, scheduled digests and reports) | Account Data (recipient names and email addresses); Customer Data contained in notification content | USA (infrastructure on AWS) | EU SCCs + UK Addendum |
| Slack Technologies, LLC (a Salesforce company) | Message delivery for the optional Slack channel. Used only where an Organization admin has installed the Fieldwork Slack app | Customer Data in messages and replies; Slack workspace, channel and user identifiers | USA (messages in transit; message history is retained in the customer's own Slack workspace under the customer's own agreement with Slack) | EU SCCs + UK Addendum; EU–US DPF |
| Meta Platforms, Inc. (WhatsApp Business Platform) | Message delivery for the optional WhatsApp channel. Used only where a User has connected their own WhatsApp number | Customer Data in messages and replies; the connecting User's phone number | USA (messages in transit; Meta retains delivery metadata) | EU SCCs + UK Addendum; EU–US DPF |
Processing locations
For models served through Vercel AI Gateway, downstream inference providers may also include xAI, Fireworks AI, Baseten, Modal, Together AI, DigitalOcean, Wafer, and Nebius, depending on the selected model and our configured provider allowlist. They receive Customer Data included in model requests and associated Usage Data for inference. The training, retention, and international-transfer safeguards described on this page apply to this processing as well.
Data at rest remains in the locations listed above. AI inference requests are processed globally by default unless otherwise agreed: requests may be routed to, and processed transiently in, any region in which the providers listed above operate their commercial services. Inference is not routed to any jurisdiction subject to comprehensive US, UK, or EU sanctions or export restrictions. Customers with data-residency requirements may request geography-pinned inference (US or EU) as a per-Organization configuration.
Training and retention
All AI model providers are engaged on terms that prohibit training their models on Customer Data, and Fieldwork likewise does not train foundation models or large language models on Customer Data. Inference requests are configured for zero or minimal retention where the provider and model support it; certain models require limited retention by the model provider for trust-and-safety purposes, and such retained data is not used for model training.
Notes
Not every sub-processor listed processes every customer's data; routing depends on configuration and region. Customers with negotiated sub-processor consent rights (rather than notice) are governed by their agreement, and additions take effect for them per that agreement.