Subprocessors
To run Integrity, we share limited data with the trusted third-party services below ("subprocessors"). This page lists each one, what it's used for, the data it may process, and where it operates. We keep it current and notify Team/business customers of material changes per our Data Processing Agreement.
Infrastructure & platform
| Subprocessor | Purpose | Data processed | Region |
|---|---|---|---|
| Hetzner Online GmbH | Cloud hosting & compute (application servers, PostgreSQL, Redis) | All account & workspace data, at rest and in transit | Germany & Finland (EU) |
| Cloudflare, Inc. (R2) | Object storage for file uploads; DNS / CDN | Uploaded files & attachments; request metadata | United States / global |
| Amazon Web Services, Inc. (SES) | Transactional email delivery | Recipient email address & message content | EU (Frankfurt, eu-central-1) |
| Sentry (Functional Software, Inc.) | Error & performance monitoring | Diagnostic/error data, IP, technical metadata | EU (Germany) |
Payments
| Subprocessor | Purpose | Data processed | Region |
|---|---|---|---|
| Stripe, Inc. | Payment processing & subscription billing | Name, email, billing details, payment metadata (card data handled directly by Stripe) | United States |
Authentication, analytics & feedback
| Subprocessor | Purpose | Data processed | Region |
|---|---|---|---|
| Google LLC (OAuth) | "Sign in with Google" | Account identifiers from Google | United States |
| Google LLC (Analytics) | Aggregated, IP-anonymized usage analytics | Anonymized usage & device data | United States |
| Userback Pty Ltd | In-app feedback & bug reports | Feedback content, contact info, technical metadata | Brisbane, Australia |
AI model providers
AI providers are engaged only when you use an AI feature with the corresponding model. Your prompt and the context you submit are sent to the selected provider to generate a response. We do not allow these providers to train on your private content.
| Subprocessor | Purpose | Data processed | Region |
|---|---|---|---|
| Anthropic, PBC | AI model (Claude) | Prompt & submitted context | United States |
| OpenAI, L.L.C. | AI model (GPT) | Prompt & submitted context | United States |
| Google LLC (Gemini API, Vertex AI) | AI model (Gemini) | Prompt & submitted context | United States |
| Nebius B.V. | AI inference | Prompt & submitted context | Netherlands (EU) |
| fal — Features & Labels, Inc. | AI image generation | Image prompts | United States |
| Gonka (OpenGNK / proxy.gonka.gg) | AI inference on a decentralized GPU network | Prompt & submitted context | Distributed / global |
| TypeSafe AI, Inc. | AI typed decisions (routing, memory classification & where new content is placed) | Recent messages, submitted context & the titles of the pages being written to | United States |
| Fireworks AI, Inc. | AI inference for open models (DeepSeek), and document reading | Prompt, submitted context & uploaded document text | North America, Europe, Asia-Pacific |
| Deepgram, Inc. | Speech-to-text for meetings and voice | Meeting & voice-message audio | United States |
| xAI Corp. | AI model (Grok) | Prompt & submitted context | United States |
Fireworks runs open-weight models — DeepSeek V4.1 Flash
today — on its own infrastructure (North America, Europe and Asia-Pacific; none
in China) and keeps no prompt data. It also
reads uploaded documents whenever the model you picked cannot read them
itself — Claude and Gemini models read their own, OpenAI and Grok
models do not, so on those your document text is sent to Fireworks.
Gonka is a decentralized AI compute network — inference
runs on independent GPU hosts that may be located anywhere in the world.
If you need to restrict which providers your workspace uses, contact
[email protected].
Entities and regions verified June 2026. Update this page whenever a
provider is added or removed in the codebase/infra.