Automatic PII detection and redaction in every request. Regex patterns catch structured data in under 1ms. Local AI models handle names, addresses, and context-dependent entities.
Grepture scans every outbound request for sensitive data — emails, credit cards, API keys, names, addresses, and more — and redacts them before the request leaves your infrastructure. On the return path, redacted tokens are rehydrated so your app gets complete responses.
Detection rules run in the proxy layer. Free tier uses 50+ regex patterns for structured PII. Pro adds locally-hosted NER models for context-dependent entities like names and organizations, plus prompt injection detection. All processing happens on Grepture infrastructure — no data sent to external AI services. A dedicated POST /v1/embeddings endpoint runs the same detection pipeline on embedding inputs so PII never ends up in your vector store — see the use case for details.
How an AI aggregator uses Grepture to redact API keys, customer records, and stack traces from every user prompt — provider-agnostic, with reversible tokens.
Read the case study →npm install @grepture/sdk — wrap your client — done.
Free for up to 1,000 requests/month · No credit card required
Get Started Free