Legal
Subprocessors
Last updated: 2026-08-27. These are subprocessors Docsiv engages to operate the Service. The list may change; see the Data Processing Agreement for how we notify customers. Questions: hello@docsiv.com.
When you use Docsiv, we may use the vendors below to process personal data on our behalf. Their use is subject to our agreements with them and their published privacy notices. For how we handle personal data overall, read our Privacy Policy.
| Subprocessor | Purpose | Typical region | Privacy |
|---|---|---|---|
| Supabase, Inc. | Hosted database, authentication, and file storage for accounts, workspace data, comments, and uploaded Customer Content. | United States (project region may vary; see Supabase documentation) | Notice |
| Vercel Inc. | Application hosting, edge routing, Web Application Firewall, Vercel Analytics, Speed Insights, and (where enabled) the Vercel AI Gateway used to route AI requests. | United States / global edge | Notice |
| Anthropic, PBC | AI model inference only (Claude models) when a Claude model is configured for an AI feature. Prompts and related Customer Content are processed to generate the requested output and are not used by Docsiv to train models. | United States | Notice |
| Google LLC | AI model inference only (Gemini models, including image generation, image editing, and audio transcription) via the Vercel AI Gateway; optional “Sign in with Google” authentication; and optional Google Drive import of files you select. Drive import does not send file content to AI. Workspace user data is not used to train models. | United States / global | Notice |
| OpenAI, L.L.C. | AI model inference only (GPT models) when an OpenAI model is configured for an AI feature. Prompts and related Customer Content are processed to generate the requested output. Requests set store=false so OpenAI does not persist them for model training or their dashboard. | United States | Notice |
| xAI Corp. | AI model inference only (Grok models) when a Grok model is configured for an AI feature. Prompts and related Customer Content are processed to generate the requested output and are not used by Docsiv to train models. | United States | Notice |
| Perplexity AI, Inc. | Optional web-search / research inference (Sonar models) when an AI feature looks up public web information. Used for search results, not to train Docsiv or Perplexity models on Google Workspace user data. | United States | Notice |
| Liveblocks Inc. | Optional real-time collaboration infrastructure (presence, rooms, and related APIs) for co-editing and live awareness, where enabled. | United States / EU (see Liveblocks documentation) | Notice |
| UploadThing (Ping Labs, Inc.) | Managed file uploads and temporary handling of upload sessions for Customer Content where the product uses UploadThing. | United States | Notice |
| Resend (Plus Five Five, Inc.) | Transactional and lifecycle email delivery (for example authentication, notifications, billing, signing reminders, and comment digests), including bounce and complaint handling. | United States | Notice |
| Dodo Payments | Payment processing, subscription and credit-pack billing, invoicing, and fraud prevention for paid plans. Card details are handled by Dodo; we do not store full card numbers. | Refer to Dodo Payments disclosures | Notice |
| PostHog, Inc. | Product analytics (usage events and performance), loaded only after you accept analytics cookies. Events are proxied through our domain. | United States | Notice |
| Unsplash Inc. | Optional stock-image search you initiate from the media tools; your search terms are sent to Unsplash to return results. Used only when you choose to search Unsplash. | United States | Notice |
AI features route requests through the paid Vercel AI Gateway for inference only. The model provider that receives your content depends on the model selected for that feature (Anthropic, Google Gemini, OpenAI, xAI, or Perplexity; additional Gateway families such as Meta Llama, Mistral, Moonshot Kimi, or Alibaba Qwen only if an administrator enables them). We do not use that content to train generalized or foundational AI models. Google Workspace user data imported from Drive is handled as described in our Privacy Policy (Google Workspace APIs and Limited Use). Optional third-party services such as Iconify (icon search) and embed providers (for example YouTube, Loom, Figma, and Slides.com) are only contacted when you choose to use them, and any data shared is governed by their own notices.
Additional vendors may be used for limited operational purposes (for example infrastructure diagnostics) at Docsiv’s discretion. If we add or replace a material subprocessor for customer personal data, we will update this page and provide notice consistent with our DPA.