Portkey allows workspace owners to enforce request guardrails at the workspace level. This feature ensures that all API requests made within a workspace comply with predefined policies, enhancing security, compliance, and governance at a more granular level.
Workspace owners can define input and output guardrails in the workspace settings. These guardrails are enforced on all API requests made within the workspace, ensuring uniform policy enforcement across all users and applications in that workspace.
Input Guardrails: Define checks and constraints for incoming LLM requests within the workspace.
Output Guardrails: Ensure LLM responses align with the defined guardrail policies.
The guardrails available here are the ones created within the user’s workspace. Users can select any of these as the default input/output guardrails for their workspace. Multiple providers are supported for setting up guardrails. For a detailed list of supported providers and configurations: