Built for sensitive models and confidential data.
We know that diagnostic evaluation requires trust. Here is how we protect your proprietary models and intellectual property at every layer.
Encrypted Credential Storage
All API credentials, endpoint authentication headers, and connection tokens are encrypted at rest using AES-256-GCM. Keys are decrypted strictly in memory at the moment of evaluation and never exposed in client bundles or network logs.
Zero Model Training Guarantee
Your prompt evaluations, completed inferences, and proprietary domain data are strictly your intellectual property. Tanvelo will never use your inputs or outputs to train public foundational or auxiliary AI models.
Sandboxed Network Execution
Benchmark requests are dispatched through dedicated egress gateways with configurable IP allowlisting. Responses are parsed in isolated ephemeral execution environments to prevent malicious payload execution.
No Autonomous Modification
Tanvelo is an independent diagnostic and advisory platform. We do not have write access to your code repositories, weights, or production deployment pipelines. We advise and quantify; you execute.
Configurable Retention & Immediate Purge
Organizations can configure strict evaluation log retention horizons. Delete full run histories, test transcripts, and synthesized datasets instantly with a single click or API call.
Role-Based Access Control
Granular workspace permissions powered by Clerk enterprise SSO ensure team members only access approved projects and evaluation reports.
Have specific compliance questions?
We accommodate custom Data Processing Agreements (DPA), HIPAA, and dedicated on-premise deployments for qualified enterprise customers.