Subprocessors

Last updated: 21 October 2026

Äng uses the third-party service providers (subprocessors) listed below to provide the Service. Each provider receives only the data needed for the stated purpose. Models marked “via the AI Gateway” are reached through Vercel, which forwards the request to the model provider. Location shows where we understand each provider processes the data; it is not a statement about the contract we hold with that provider.

We update this page when we add, replace, or remove a subprocessor. If you have an active Data Processing Agreement with us and want to be notified of changes, contact us at privacy@ang.studio.

Supabase

  • Purpose: Database, authentication, and file storage
  • Data processed: Account data (email, user ID), boards, images, files, and other user content
  • Location: EU

Vercel

  • Purpose: Application hosting, edge network, bot protection, AI Gateway model routing, and cookie-free web analytics (analytics only after consent)
  • Data processed: Technical request data (IP address, request logs), usage events; prompts and media for models routed through the AI Gateway (not including direct integrations such as fal.ai)
  • Location: Global edge network

Google

  • Purpose: Sign-in (Google OAuth); Gemini / Veo models for image and video generation and Gemini text-to-speech for generated voice (via the AI Gateway); automatic visual analysis of library images for AI search; and Vertex AI hosting for board agent chat in some configurations, including Zero Data Retention mode
  • Data processed: Login identifiers; prompts, voice scripts, images, and files you submit or store that are processed by Google models or Vertex-hosted agent requests
  • Location: EU/US

OpenAI

  • Purpose: GPT Image models (via the AI Gateway); realtime voice transcription (only when you use voice input); and board agent chat when the AI Gateway routes it to OpenAI
  • Data processed: Prompts and images for GPT Image jobs; audio snippets and transcripts when you use voice input; conversation text, board context, and any images included in agent requests routed to OpenAI
  • Location: US

Microsoft

  • Purpose: Azure OpenAI hosting — one of the providers the AI Gateway may select for board agent chat. Under strict Zero Data Retention, agent requests are routed only to providers operating under zero-data-retention agreements, and refused if none is available. Also MAI Image models on Azure (via the AI Gateway), when you choose one
  • Data processed: Conversation text, board context, and any images included in agent requests routed to Azure OpenAI; prompts and images for MAI Image jobs
  • Location: US/EU for agent chat; global, via the AI Gateway, for MAI Image

Anthropic

  • Purpose: Board agent chat and related text models in some configurations (via the AI Gateway; in Zero Data Retention mode those requests are hosted on Google Vertex)
  • Data processed: Conversation text, board context, and any images included in agent requests
  • Location: US

ByteDance

  • Purpose: Seedream image and Seedance video models (via the AI Gateway), only when you select those models
  • Data processed: Prompts, images, and related inputs for those jobs
  • Location: US/global, via the AI Gateway

xAI

  • Purpose: Grok Imagine image models (via the AI Gateway), only when you select those models
  • Data processed: Prompts, images, and related inputs for those jobs
  • Location: Via the AI Gateway

Alibaba

  • Purpose: Wan video models (via the AI Gateway), only when you select those models
  • Data processed: Prompts, start-frame images, and related inputs for those jobs
  • Location: Via the AI Gateway

Meta

  • Purpose: Muse Image models (via the AI Gateway), only when you select those models
  • Data processed: Prompts, images, and related inputs for those jobs
  • Location: Via the AI Gateway

Black Forest Labs

  • Purpose: FLUX image models (via the AI Gateway) and the FLUX 3 Max video model (via fal.ai), only when you select those models
  • Data processed: Prompts, images, and related inputs for those jobs
  • Location: Via the AI Gateway or fal.ai

fal.ai

  • Purpose: Direct API integration (not the AI Gateway) that hosts or relays specialist models — only when you use these features, which require Standard mode: image-to-3D (Tencent Hunyuan 3D, Meshy, Tripo), Change material on 3D models (Meshy), Split into layers (ByteDance Seedream), PBR material extraction, Smart Select (Meta SAM 3), Vectorize (Recraft), camera-angle variants, Ideogram and Krea image models, and the FLUX 3 Max video model
  • Data processed: Images, 3D models, and prompts you submit to those features, and the results generated from them
  • Location: US

MiniMax

  • Purpose: MiniMax video models (via the AI Gateway), only when you select those models
  • Data processed: Prompts, start-frame images, and related inputs for those jobs
  • Location: Via the AI Gateway

Ideogram

  • Purpose: Ideogram image models (via fal.ai), only when you select those models
  • Data processed: Prompts, images, and related inputs for those jobs
  • Location: Via fal.ai

Krea

  • Purpose: Krea image models (via fal.ai), only when you select those models
  • Data processed: Prompts, images, and related inputs for those jobs
  • Location: Via fal.ai

Topaz Labs

  • Purpose: Video upscaling (via the AI Gateway), only when you use Upscale on a video
  • Data processed: The video you upscale and its upscaled result
  • Location: Via the AI Gateway

Meshy

  • Purpose: Image-to-3D generation and Change material on 3D models (via fal.ai), only when you use those features
  • Data processed: Images, 3D models, and material prompts you submit to those features, and the 3D models generated from them
  • Location: Via fal.ai

Tripo

  • Purpose: Image-to-3D generation (via fal.ai), only when you select a Tripo 3D model
  • Data processed: The images you submit and the 3D models generated from them
  • Location: Via fal.ai

Recraft

  • Purpose: Vectorize (via fal.ai), only when you vectorize an image
  • Data processed: The image you vectorize and the vector result
  • Location: Via fal.ai

Stripe

  • Purpose: Payment processing for subscriptions and credit top-ups (when paid features are used)
  • Data processed: Billing contact details, transaction metadata; card data is handled by Stripe, not stored by us
  • Location: US/EU

Cloudflare

  • Purpose: Bot protection on public signup
  • Data processed: Technical challenge data and IP address during verification
  • Location: Global edge network

Sentry

  • Purpose: Error monitoring and crash reporting
  • Data processed: User ID, email, device and error context; error replays with text masked and images blocked
  • Location: EU (Frankfurt)

PostHog

  • Purpose: Product analytics (only after you consent to analytics)
  • Data processed: User ID (UUID only — never email or name), feature-usage events
  • Location: EU (Frankfurt)

Upstash

  • Purpose: Rate limiting and short-lived caching
  • Data processed: User identifiers and request counters; short-lived signed links to your stored files
  • Location: Per deployment region

Resend

  • Purpose: Transactional email (sign-in links, invites, lifecycle messages)
  • Data processed: Email address, message content and delivery metadata
  • Location: US

Notion

  • Purpose: Internal management of in-app feedback (only when you submit feedback)
  • Data processed: Feedback message, your email address
  • Location: US

Content delivery (no user content)

Some on-device features (such as background removal) download machine-learning model files from the jsDelivr and Hugging Face content delivery networks. Your browser's IP address is visible to these CDNs when fetching the files, but no user content or account data is sent to them.

See our Privacy Policy for how we handle personal data, our Data Processing Agreement for business customers, and our Cookie Policy for cookies and similar technologies.

← Back to sign in