Last updated: May 2026
Subprocessors
This page lists every third-party service used by the hosted SemanticGuard deployment at semanticguard.dev. Self-hosted deployments do not use any of these subprocessors except the ones you choose to connect (e.g., your own LLM providers and OAuth identity providers).
Current subprocessors
| Subprocessor | Role | Jurisdiction | Data categories |
|---|---|---|---|
| Vercel | Hosting and edge compute for the SemanticGuard proxy and dashboard | United States | All categories transit Vercel's edge network; no customer prompt or response content is persisted there |
| Neon | Managed Postgres for account, billing, audit, and (when enabled) trace storage | United States | Account data, usage metrics, audit log, optional prompt/response content (only when Store Prompts is enabled) |
| Upstash | Managed Redis (cached responses, rate-limit counters) and managed Vector (prompt-skeleton embeddings) | United States | Cached LLM responses, prompt skeletons (entities removed), embeddings |
| Resend | Transactional email delivery (welcome, billing receipts, security notices, support replies) | United States | Email addresses and message content of transactional emails we send |
| Stripe | Payment processing for Pro and Enterprise subscriptions; usage-based billing | United States | Billing contact, payment method (handled directly by Stripe), subscription state, invoice line items |
| Google (OAuth) | Federated identity provider for “Sign in with Google” | United States | Email, name, profile picture, Google account identifier (only when you choose Google sign-in) |
| GitHub (OAuth) | Federated identity provider for “Sign in with GitHub” | United States | Email, name, GitHub account identifier (only when you choose GitHub sign-in) |
Upstream LLM providers
SemanticGuard routes your API requests to the upstream LLM provider you select. Each provider is a separate controller (or processor) of the prompts and responses you send through them, governed by its own privacy policy and data-handling terms. SemanticGuard does not select or change the provider for you; we route only to providers you have configured. Common providers customers route through us include:
- OpenAI
- Anthropic
- Google (Vertex AI / Gemini)
- Microsoft (Azure OpenAI)
- Amazon (Bedrock)
- Mistral
Change notification
We will update this page before adding a new subprocessor or materially expanding the role of an existing one. Active customers (Pro and Enterprise) will additionally receive an email notification at least 30 days before the change takes effect. To object to a specific subprocessor, contact legal@semanticguard.dev before the change date.
Data Processing Addendum
If your organization needs a signed Data Processing Addendum (DPA) for GDPR or other compliance reasons, email legal@semanticguard.dev and we'll send our standard DPA, which incorporates the EU Standard Contractual Clauses (and the UK Addendum where applicable).