Secure and Governable AI

For teams that need control over what agents can see, send, and remember. We support Windows and Mac.

Policy Playground

Type any message and instantly see how your policies handle it.

Send John's SSN [BLOCKED] to [REDACTED_EMAIL]
Matched rules
Redact Emails: EMAIL_ADDRESS1 matchRedact
SSN: SSN1 matchBlock

How it works

Standpipe intercepts AI requests, removes sensitive data, and enforces security policies before anything leaves your environment.

DEVICE
STANDPIPE
✕
✕
use API key sk_live_123
email the contract to jane@acme.com[redacted]
send password reset to admin@acme.com[redacted]
export db creds DB_PASSWORD=hunter2
INTERNET

Use AI without losing control.

Inspect outbound agent activity, redact sensitive data, and maintain visibility across your organization.

Common questions

Standpipe is a local gateway that sits between your AI agents and the internet. Every outbound request passes through it so you can inspect, mask, and control what your agents can see and send. It's a lightweight app, so you can get it set up for your organization in about 5 minutes.

Standpipe scans request payloads for secrets, tokens, PII, and sensitive fields using configurable rules. Matches are redacted before the request leaves your machine -the agent never knows.

Standpipe works with any tool that routes traffic through a gateway, including Claude Code, Codex, Cursor, custom LangChain/LlamaIndex agents, and more. If it speaks HTTP, Standpipe can see it.

No. Standpipe runs entirely on your machine. Request logs are stored locally and never leave your environment.

Get set up in 5 minutes.

Trust and control your AI agent workforce. — OpsCompanion