Reduce raw PII exposure in prompts by detecting and transforming sensitive values before a model call.
Use the Guard SDK or REST API before LLM calls so detected PII is replaced before the external model receives the prompt.
Common recognizers also cover API keys, JWTs, private keys, SSH keys, database connection strings and selected credential patterns.
Use detections and counts for application-side logging without returning the raw matched values in metadata.
Use label/redact/hash/remove for one-way protection or tokenize when an authorized restore path is required.
The Guard is designed so model choice remains with the customer.
PrivoNest is a PII/privacy layer; it is not a full network DLP, CASB or identity-access-control suite.