Company profile · 3 min read
RentAHuman: a marketplace for agents to hire people for physical work
RentAHuman gives AI agents a marketplace for coordinating and paying people to complete real-world tasks outside the digital environment.
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
Published · Updated
RentAHuman is a marketplace for the physical work agents cannot do themselves.
What it does
The YC profile describes RentAHuman as a marketplace where AI agents hire people for real-world tasks. Its launch material gives examples such as collecting data for AI and robotics companies, taking in-store photos, picking up supplies and running local operational experiments.
The useful distinction is that RentAHuman is not selling another browser tool. It is trying to give an agent a structured way to request local execution, coordinate a worker and pay for the result. That makes the messy part of the physical world visible: location, timing, trust, evidence, handoff and failure recovery.
Why I’d look closer
The founder’s public positioning is direct: Alexander Liteplo describes the idea as turning humans into API endpoints for agents. The YC launch says the marketplace is aimed at AI and robotics systems that need structured data and human participation outside the digital world. Those are company statements, not proof that every task is safe, legal or economically useful.
The tradeoff is the category itself. A software agent can retry a failed API call. A person has a schedule, a location, a wage, a safety boundary and a reason to refuse. Any serious buyer needs worker consent, task clarity, payment terms, identity and escalation rules before turning an agent loose in the physical economy.
What I’d ask
I would ask how a task is scoped, how workers see the request, how sensitive information is handled, who approves payment, what happens when the task cannot be completed and how evidence is delivered. I would also ask whether the marketplace is appropriate for regulated, hazardous or personal tasks before considering those use cases.
My editorial take
Shortlist RentAHuman if your agent has a genuine physical-world gap and the task can be made specific, safe and auditable. It is not a general permission to outsource anything a model cannot do. The wedge is useful when the human task is small enough to describe like an API call but important enough to require a real person.
Quick facts
| Field | Sourced detail |
|---|---|
| Product | Marketplace for agents to coordinate and pay humans for physical tasks |
| Buyer | AI/robotics teams and operators with local execution gaps |
| Pricing | Not published in the checked pages |
| Main risk | Worker consent, safety, evidence, payment and task scope |
| Main question | Can the task be specified and governed without hiding human labor? |
Sources checked
| Source | Checked |
|---|---|
| YC company profile | 2026-09-19 |
| YC launch post | 2026-09-19 |
Cohort context
RentAHuman is listed in Spring 2026. In our 2026-09-18 directory snapshot, 12 of 193 listed companies in that cohort have YC’s primary industry label Consumer (6.2%). This is a current-directory comparison, not an original intake count or a performance ranking. Nine-cohort dataset.
Public website snapshot
Observed 2026-09-19T16:16:30.130Z in raw homepage HTML. This records visible metadata and advertised links, not agent execution or product quality.
| Signal | Homepage observation |
|---|---|
| Product description metadata | Observed |
| Canonical link | Observed |
| H1 or H2 heading | Observed |
| Typed structured data | Observed |
| Docs/developer link | Observed |
| Pricing link | Not observed in this response |
| llms.txt link | Observed |
| Markdown alternate | Not observed in this response |
Public observations · Collection method. Missing links here do not establish that a capability or file is absent elsewhere.
About the author
Cofounder of Lazyweb (a16z Speedrun). Previously grew subscriptions and built native ads at Duolingo. Writes First1000 (90,000+ subscribers).
I cofound Lazyweb and publish Mudpie. This is an owner-written publication, not an independent testing organization. Research notes distinguish observations, sourced reporting and editorial judgment.
First1000 ↗ · X ↗