Euclid provides an Anthropic-compatible API that routes inference to Amazon Bedrock in European Union / EEA regions. This notice explains what we collect and what we deliberately do not — including how Euclid supports internal policies for code-generating AI tools.
Policy alignment (code-generating AI)
- DPA: A Data Processing Agreement is available for customers who need a processor contract before approving tooling.
- EU/EEA processing: Model inference uses AWS Bedrock EU geographic profiles so processing remains within the EU/EEA.
- No training on customer data: Euclid does not store prompts or completions, and does not use your content to train models. Inference is provided for your use of the service, not for model improvement on your data.
What we store
- Account details needed to operate your organization (name, email).
- Payment and prepaid credit ledger data (via Stripe when configured).
- Usage metadata: request id, API key id, model, token counts, cost, latency, and status.
Workspaces
A private workspace belongs to one person. A company workspace shares that workspace's usage metadata and prepaid balance with its members. An invitation stores the invited email address until it is accepted, revoked, or expires. A private workspace stays separate from any company workspace on the same account.
What we do not store
- Prompt contents
- Model completions / responses
- Source code or chat bodies beyond transient proxying of a live request
Inference location
Model inference is performed using AWS Bedrock EU geographic inference profiles so processing remains within EU regions. Euclid's control plane aims to keep customer and billing data in the EU as well.
DPA & contact
Request a Data Processing Agreement or privacy follow-up at [email protected].