TETRA inspects AI prompts in real time. Your company decides which data needs control and what action to take when it is detected.
An employee writes a prompt in a supported AI service.
The TETRA extension or agent checks the content on the employee’s device using configurable rules and local detectors.
Policy determines the action: monitor, mask, or block.
The original or masked prompt is sent to the AI service, or the prompt is stopped.
Monitor, mask, and block are alternative actions, not a sequence.
The prompt is sent unchanged. The security team sees which AI services employees use and which data categories were detected. Helps assess activity before restrictions, but on its own does not prevent sensitive data from being sent.
TETRA hides detected sensitive information according to company rules. Masking is irreversible: data is not replaced with tokens for later recovery. The modified prompt is sent, and the response may differ without the hidden details.
TETRA stops prompts that violate company rules. When a prompt is blocked or masked, the employee sees a notification, and the customer controls its text.
In the demo, we’ll show how a prompt is inspected and the resulting event in the administrative console.
