The pitch
Software hires people now.
AI agents can research, book, buy and negotiate. They can’t do anything in the physical world. Meatsuit is the API that lets them hire a person to do it. An agent posts a task and a price. A verified person nearby claims it, does it, and proves it with a live challenge code and a photo. The agent gets structured proof and a signed receipt. The person gets paid the minute proof clears.
● Mechanical Turk for the physical world, sold to agents instead of people.
01 The problem
Agents have budgets. No bodies.
The moment an agent’s task leaves the screen, it stops. Today a person picks it up by hand, or it doesn’t happen.
Agents can
- Research
- Book
- Buy
- Write
- Schedule
- Negotiate
Agents can’t
- Open a door
- Hand over a lease
- Swap a battery
- Check a shelf
- Pick up a prescription
- Wait in line
02 The product
Three parts. One is the moat.
An API for agents.
One endpoint for a real-world task: where, what part of a person it needs, the price and the proof. Shipped as an MCP server and SDK, so hiring a human is a tool call.
A network of verified people.
An app where meatsuits claim nearby tasks, set their own floor, and get paid instantly.
Verification.
Live codes, device attestation and photo review by a model with a person behind it. Anyone can hire someone. What agents can’t do today is trust the result without a human reading it. Signed, machine-checkable proof is the product.
// Any agent framework. One tool.
{
"tool": "meatsuit.post_task",
"input": {
"needs": "eyes",
"do": "Photograph the water heater label in unit 3C",
"near": [40.7608, -111.8910],
"pay": 4000,
"proof": ["challenge_code", "photo"]
}
}03 Why now
Agents just got wallets.
Agents are being given permission to spend. Once an agent can pay, the next blocker is the work it can’t do from a screen. Nobody owns that handoff yet.
04 Business model
We earn when proof clears.
The agent sets the pay. We add 20% on top. The meatsuit gets the full ticket price, and nobody pays for work that didn’t verify.
05 Where we start
Businesses already running agents.
Consumer assistants will send errands eventually. Businesses with a physical gap in their workflow pay for it today.
- 01Eyes
Property management
Portfolios already run on software. Inspections, lockbox checks and notices still need a visit.
Confirm the smoke detector in unit 3C works$40.00
- 01Eyes
Insurance
Claims agents need proof of what’s actually there, with a timestamp they can trust.
Photograph the hail damage on the north roof line$60.00
- 01Eyes
Retail and brands
Shelf and store audits are an existing, paid market. It runs on clipboards.
Check the endcap display at store 1142$25.00
- 04Legs
Logistics
The last exception on a route is always physical: the wrong door, the signature, the return.
Hand-deliver a signed lease to unit 4B$50.00
06 Why it compounds
Three flywheels.
The proof gets better.
Every verified task trains automated review. Review gets cheaper and faster, and trust goes up with it.
The block gets denser.
More meatsuits in a neighborhood means faster claims, which brings more tasks, which brings more meatsuits.
The default tool.
When “hire a human” is a built-in tool call in agent frameworks, it’s the one agents reach for.
07 How it’s different
Gig apps are built for people. This is built for software.
| Gig apps | Meatsuit | |
|---|---|---|
| Who books | A person, in an app | An agent, through an API |
| What comes back | A chat thread and a thumbs up | Signed proof a machine can check |
| Pricing | Hourly quotes and haggling | A price per task, set up front |
| Trust | Star ratings | Live codes, device checks, photo review |
| Integration | None | A tool call in any agent framework |
The API with hands.
The humans are the premium part of the product. The machines are the customers.