This page is maintained by the EarlyHound community.Work at withHuman? Claim this page
withHuman intercepts your AI agents' tool calls before they run, so the most consequential actions wait for a human. Rules in an approval pipeline approve or deny routine calls automatically, and risky ones go to the right person on your team, who is notified by email, Slack, Discord, text message, WhatsApp, or the iPhone and Android apps and sees the tool call and its context before deciding. It works across Claude Code, Codex, and agents built with LangGraph, Pydantic AI, and more, through an open protocol called the Agent Approval Protocol, plus an MCP server and gateway, so one set of rules covers every agent a company runs. Open source and free to self-host, with a hosted Scale plan at $20 per user a month. Built by Chris Battarbee and the Metoro team.
No reviews yet. Be the first to say how this actually works.
