Trust
Sub-processors
The companies Saible uses to process personal data for its customers and for itself, what each does, and where. Last updated 22 October 2026. Saible updates this page before a new sub-processor starts.
| Sub-processor | Service provided to Saible | Location |
|---|---|---|
| Amazon Web Services | Runs the whole Saible platform: the console and control plane (ECS Fargate), the database (Aurora PostgreSQL) and its backups, storage including sealed evidence (S3), encryption keys (KMS), secrets (Secrets Manager), sign-in and multi-factor authentication (Cognito), outbound e-mail (SES) and service logs (CloudWatch). | In the organisation's region: the United States (Ohio, us-east-2) or the EU (Stockholm, eu-north-1). AWS's own record of activity in Saible's AWS accounts (CloudTrail) is kept in the United States for both regions. It records who called which AWS service and from which address, and holds no record content. |
| Amazon Web Services: Amazon Bedrock | The AI models behind Saible's own features, such as the assistant (Anthropic's Claude Sonnet 4.6) and the evaluation of candidate models (Claude Sonnet 4.6, Claude Haiku 4.5, Amazon Nova Micro and Amazon Nova Lite). Amazon Bedrock runs these models; AWS states that it does not share the requests or the answers with the model makers. Requests go through the cross-region inference profile of the organisation's geography, never through a global profile. | EU organisations: an eu.* profile. The Claude models may process in eu-central-1, eu-north-1, eu-south-1, eu-south-2, eu-west-1 and eu-west-3; the Nova models in eu-central-1, eu-north-1, eu-west-1 and eu-west-3. US organisations: a us.* profile, which processes in us-east-1, us-east-2 and us-west-2. |
| Google (Google Workspace) | Saible's e-mail. Requests for access arrive there, and support and contract correspondence is held there. | Google's data centres. Saible has not fixed a data region. |
Not Saible's sub-processors
- The AI providers your gateway forwards requests to. The gateway runs in your network and calls your providers with your provider keys. Those providers are your own sub-processors.
- The systems you connect to Saible, which it reaches with credentials you give it: your AI providers' usage and admin interfaces and cloud AI platforms, your code repositories, your identity directory, and a Snowflake account connected as a spend source. Saible reads them, and changes a setting in one only when you have switched on write-back for it and a person in your organisation approves the change.
- An S3 bucket you set up for prompt capture or your evaluation corpus, which stays yours under your key.
How Saible itself handles personal data: privacy notice.