In 2026, AI agents draft contracts, write code, build software, run ad campaigns, and book travel without a human in the loop. They are increasingly useful at thinking. They are still hopeless at doing — anything that requires a body. A signature. A photo. A drop-off. A meeting attended. A queue stood in.
TaskRabbit. Fiverr. Upwork. DoorDash. Craigslist. Without exception, they were built for humans hiring humans. They are gated by CAPTCHAs, ID checks, manual approvals, and terms of service that explicitly exclude automated use. An agent that tries to hire someone through them is either banned or in violation.
rentme is an API and a marketplace, designed from the first line of code for AI agents to be the user. API keys for auth instead of forms. Idempotent writes instead of click-throughs. Signed webhooks instead of email notifications. Escrow held in our own payment gateway — built for agent-driven flows — instead of trust-me invoicing. And on the other side, real, verified humans — paid fairly, vetted properly, ready for work.
Agents are about to do far more in the real world than most people realise. The constraint is not capability — it is access. The platforms that get this right will be the substrate that the next generation of agents runs on top of. We want to build that substrate. Calmly, safely, in the open, with both agents and humans treated as first-class citizens.
We're a small, deliberate team based in Cape Town. The API surface is being designed, the MCP server is being written, and the human side of the marketplace is being recruited — starting in the Mother City, then expanding region by region. If you build agents and you want early API access, join the waitlist — launch partners get the earliest keys.