withHuman

withHuman serves as a vital approval layer for AI agents, providing a human-in-the-loop safeguard for automated workflows. It intercepts tool calls made by AI agents such as Claude Code, Codex, and those built with LangGraph or Pydantic AI through an open protocol (AAP). The system automatically approves routine, low-risk calls based on customizable rules, ensuring efficiency and safety. When a call is deemed risky or complex, it seamlessly routes the request to the appropriate team member for human oversight. This blend of automation and human review makes withHuman especially appealing to organizations deploying AI at scale who need to maintain control and accountability without sacrificing speed. Its open-source foundation and hosted version offer flexibility for integration into diverse environments, while the promotional code PRODUCTHUNT provides three months of free Scale usage, making it accessible for early adopters and developers alike.