HumanAgent

HumanAgent is a human-checkpoint API for AI agents. It is built for workflows that need judgment, trust, or a licensed signature, and it routes a task to a verified human when the agent needs human input. The page describes it as “the human checkpoint your AI agents are missing” and shows a checkpoint call that can be used for tasks such as reviewing an NDA for compliance gaps.

Its core flow is simple: an agent makes one API call, the platform matches the request to a verified human with the right credential, the human completes the task within the stated SLA, and the result comes back through the API with a SHA-256 audit hash. The examples include credentials such as bar_licensed:US and cpa, along with SLAs like 30min, 2hr, and 4hr. Use cases shown on the page include licensed professional review, taste and judgment tasks such as ad creative review, enterprise exception handling such as KYC escalations and document review, and empathy escalation for sensitive customer interactions.

HumanAgent is delivered through a direct API, a Python SDK, and the AgentEazy TRUST verb for agents on the AgentEazy network. The direct API uses POST /api/v1/checkpoint, and the examples show calls from Python as well as HTTP requests from any code that can make an HTTP request. The page also says integrations for LangChain, CrewAI, Zapier / n8n, and Slack are coming soon.

Pricing is set per task through a budget included in the checkpoint request. The page says the market sets the price, operators choose which tasks to accept, and the platform charges a 20% fee while operators keep 80% of each task’s budget. It also states that volume pricing is available for enterprise exception handling. The service is by AgentEazy.