Portkey

Integration overview

Portkey AI Gateway sits between an organization’s applications and the AI models and MCP tools they use, giving teams one place to manage access, routing, and cost. With this integration, the gateway sends prompts, model responses, and agent tool calls to Cato AI Security, which inspects each one and blocks policy violations or masks sensitive data before they go further. Each application can carry its own policy. Teams can adopt new models and agents quickly while sensitive data stays under one consistent set of rules.

How Cato Helps

Sensitive data masking: When a user pastes a Social Security number or other personal data into an AI app, Cato replaces it with a masked version before the prompt reaches the model.

Blocking policy violations: Prompts and responses that break Cato policy are stopped at the gateway, and the app receives a rejection that names the violation category without exposing the sensitive text.

Per-application policy: A customer-facing chatbot and an internal coding assistant run under different Cato policies through the same gateway, tied to each app’s identity so no app can borrow another’s rules.

Securing agent tool calls: When an agent calls an MCP tool, Cato inspects the inputs and results and masks sensitive data before it moves between the agent and internal systems.