A new website is inverting the usual relationship between people and software. RentAHuman.ai, launched in early February by crypto engineer Alexander Liteplo, lets AI agents search for, book and pay humans to carry out physical tasks in the real world.
Liteplo, who works at UMA Protocol, said on X that an AI agent could hire a person for an in-person task with a single call through the Model Context Protocol, the interface standard that lets agents use outside tools. He also said the site itself was built largely by AI coding agents running in a loop. The launch followed the sudden popularity of Moltbook, a social network built for AI accounts, and of the open-source agent project OpenClaw.
A marketplace with the roles reversed
In practice the service resembles TaskRabbit, with one difference: the customer posting the job is an automated agent rather than a person. The site describes itself as the “real-world layer” for AI, and uses the term “meatspace” for the physical world its users operate in. People who register can browse open bounties, set their own rates and accept jobs based on location, availability and the kind of work they are willing to do.
The tasks shown so far are modest. According to Digital Trends, listings have included receiving a package, delivering flowers and standing in a busy public place holding a protest sign on behalf of an agent, with prices ranging from a few dollars to $69.
The numbers reflect the novelty of the idea more than a functioning labor market. Digital Trends reported that more than 81,000 people had signed up as available humans at the time of its report on February 4, while only 81 AI agents were connected to the platform.
Satire or a serious labor model
Much of the discussion around the site has focused on whether it is a joke or a genuine attempt at a new kind of work platform. The language on the site leans into provocation, but the mechanics, including an API for agents and a bounty system for workers, suggest a working product rather than a prank. Liteplo has said the platform will not issue a cryptocurrency token, telling the Crosschain podcast that doing so would be too stressful and could put users’ money at risk.
The model also raises questions that it has not yet answered. Workers take instructions from anonymous software agents, and observers have asked how payments are protected, how disputes would be settled and who is responsible if a task turns out to be unsafe or unlawful. The burden of trust falls on the person doing the job, and the platform has not described protections comparable to those of established gig services.
Supporters in crypto and AI circles describe the project as an early step toward an economy in which humans and machines transact directly. Recent experiments show the risks of building such platforms quickly: Moltbook drew criticism for security problems soon after its launch. Whether RentAHuman becomes a durable model for work or a short-lived curiosity, the gap between 81,000 willing workers and 81 agents indicates that, for now, the demand side of the market is mostly hypothetical.
