Subprocessors (Public List)

Last updated on November 3, 2026

Contact: supportdocumentation.ai

This page discloses third‑party service providers that process personal data on behalf of Keevs Health Inc. (Documentation.AI) to deliver, secure, and support our Services. We require each subprocessor to use personal data only to provide contracted services, implement appropriate security measures, and comply with applicable law. This list forms Annex III of our Data Processing Addendum.

Change management: We will update this page and provide at least 30 days’ prior notice before adding or replacing a subprocessor, except for emergency replacements (we will notify promptly thereafter). To raise an objection where permitted by contract, email supportdocumentation.ai with “Subprocessor Objection” in the subject.

Current subprocessors

VendorPurposeCategories of dataData location(s)Key safeguards
Render Services, Inc.Application hosting: API, background workers, AI assistant, MCP server, PDF renderingAll categories processed by the Services, in transit and in memory; application logsUnited States (Oregon, Ohio)TLS; network isolation; access controls
Vercel, Inc.Hosting of the dashboard and published documentation sites; edge network; AI Gateway routing to model providersCustomer Content; reader request data (IP address, request headers); AI requests in transitUnited States; global edge networkTLS; encryption at rest; isolation; access controls
Supabase, Inc.Managed PostgreSQL databaseAccount and workspace data (names, emails, roles); documentation metadata; encrypted integration credentialsUnited StatesEncryption at rest and in transit; role-based access
Cloudflare, Inc.Object storage (R2) for documentation files, images, videos, and exports; CDN and DDoS protectionCustomer Content files, uploaded images and videos, imported sources, and exports; request data (IP address, headers)United States; global edge networkEncryption at rest and in transit; DDoS mitigation
imgix (Zebrafish Labs, Inc.)Image optimization and delivery on published documentation sitesImages in Customer Content; reader request dataUnited States; global edge networkTLS; access controls
Upstash, Inc.Redis caching for site configuration, reader sessions, and AI agent contextSite configuration; reader session data (for authenticated docs: identity claims such as first name, company, access roles); agent contextUnited StatesEncryption at rest and in transit; access controls
GitHub, Inc.Git hosting for documentation repositories managed by Documentation.AICustomer Content; commit metadata (author names, commit messages)United StatesEncryption at rest and in transit; access controls
Trigger.devBackground job orchestration (deployments, sync, AI jobs)Job payloads, which may include Customer Content and account identifiersUnited StatesTLS; access controls
Inngest, Inc.Background job orchestration (PDF export, cleanup)Job payloads, which may include Customer Content and account identifiersUnited StatesTLS; access controls
Better Stack, Inc.Application log management and monitoringApplication logs (request metadata, account and documentation identifiers, error details)United StatesTLS; access controls
turbopuffer, Inc.Vector database for semantic search and Ask AI retrievalEmbeddings and text chunks of Customer Content; page metadataUnited States (AWS us-east-1)Encryption in transit; per-organization namespaces
TinybirdBuilt-in documentation analytics, search analytics, page feedback, and Ask AI analyticsSession IDs; page paths; timestamps; referrer; user agent; viewport; search queries; feedback text; Ask AI question text and answer feedbackUnited States (AWS us-east-1)TLS; access controls
Voyage AI (MongoDB, Inc.)Text embeddings for semantic search and Ask AICustomer Content and search or Ask AI query textUnited StatesTLS; no training on customer data
OpenAILarge language models for the AI documentation agent and Ask AIPrompts, Customer Content, Ask AI messages, and model outputsUnited StatesTLS; API data not used for training
Anthropic, PBCLarge language models for the AI documentation agent and its connected-tool assistantsPrompts, Customer Content, content retrieved from connected tools, and model outputsUnited StatesTLS; API data not used for training
Context.dev (formerly Brand.dev)Crawling websites Customer adds as knowledge sources for the AI agent; extracting brand assets from a domainURLs and website content Customer directs us to crawl, which may include personal data published on those sitesUnited StatesTLS; access controls
Scalekit, Inc.Connection broker for third-party tools Customer connects to the AI agent (Jira, Confluence, Linear, Notion, HubSpot, Plain)OAuth tokens for connected accounts; content retrieved from connected tools, which may include names, emails, and other personal dataUnited StatesEncryption at rest and in transit; token vaulting; access controls
Clerk, Inc.Authentication and identity managementNames; emails; profile pictures; authentication factors; session and device metadataUnited StatesEncryption in transit; access controls
Stripe, Inc.Payment processing, invoicing, and usage meteringBilling name; email; payment method token; transaction and AI-credit usage metadataUnited StatesPCI DSS Level 1; tokenization (we do not store full card numbers)
IntercomCustomer support and in-app messagingName; email; user ID; support messagesUnited StatesEncryption at rest and in transit; access controls
PostHog, Inc.Product analytics in the Documentation.AI dashboardUser and organization identifiers; name; email; feature usage eventsUnited StatesTLS; access controls
LoopsTransactional email (for example, export-ready notifications)Name; email; message contentUnited StatesTLS; access controls
hCaptcha (Intuition Machines, Inc.)Bot and abuse protection on Ask AIIP address; browser and device signalsUnited StatesTLS; used only for abuse prevention

Notes

  • Integrations you enable: third-party analytics, chat, and custom scripts that you add to your documentation site send data directly to those providers under your own agreements with them; they are not our subprocessors.
  • Regionality: Where supported, we select regions to align with our hosting footprint; exact regions may change as we scale, subject to the notice above.
  • AI models: We use model providers as configured to fulfill user requests. We do not permit model providers to train on customer content.