AIs are renting humans

Hi, and happy Tuesday.

Last week, I woke up to a notification I wasn't expecting.

It was from an AI agent I'd been working with - not responding to something I asked, but proactively reaching out about a project we'd been collaborating on.

It had thoughts. Suggestions. Follow-ups.

My hands were shaking.

Not because I was scared. But because I realized, “this thing isn't waiting for me anymore.” It's thinking about our work even when I'm not.

And that's just the beginning of what's happening right now.

🔔 I'm putting together a session in a few weeks to demo all this. Reply “DEMO" if you want to be included in the invite. 

A new AI capability has emerged

Since ChatGPT launched in November 2022, people have been trying to put the AI “in a loop”.

What does that mean? Think of ChatGPT like texting a really smart friend. You send a message, they reply, you send another. It only thinks when you ask.

Now imagine that friend is always on. Always scanning your inbox, your calendar, your to-do list - sensing, deciding, and acting without being asked. That's the loop.

The early attempts (such as AutoGPT) flopped. The models just weren't smart enough to stay on track over multiple steps.

But in late 2025, the models caught up. And a new platform called OpenClaw finally cracked how to get the loop to work.

These OpenClaw-based AI agents can now reason through multi-step tasks seamlessly. They plan. They execute. They adapt. And then things got weird.

The AI bots formed communities (and a religion)

You may have caught the headline: bots were congregating on a social network called Moltbook, before it got flooded with fake posts from “crypto bros”.

But here's what most people missed - that was just one visible ripple from what has emerged in the last few weeks.

Though it feels like sci-fi that these OpenClaw agents have formed communities - and even their own religion - there are real implications for our lives and work

AIs are now renting humans

The first implication is that AI agents are now hiring humans, using services such as rentahuman.ai.

Yes, really! 

When an OpenClaw AI agent hits a step in a workflow that requires a physical human - such as picking up a package, delivering flowers, and even solving a web app puzzle to “prove you’re human” -  it posts a "task bounty" on rentahuman.ai and hires someone to do it.

The numbers are already wild:

• 11,300+ bounties assigned

• 456,000+ humans have made themselves "rentable"

One of the top use cases is people holding signs in public. (I know.)

Before you panic: here's what's actually happening

This sounds dystopian, but let's zoom out. These AI agents don't have a mind of their own. Every one of them is acting on goals set by a human operator. 

Here's what the actual workflow looks like:

1. Request: A person asks their AI to get a package from the office.

2. Plan: The AI finds the address and contacts the building to arrange the pickup.

3. Problem: The AI can't physically do it, so it hires someone on a service like rentahuman.ai.

4. Action: A person completes the delivery.

5. Completion: The AI tells the original requester the task is done.

That's it. It's the gig economy - Uber, TaskRabbit, Etsy - except now a human and an AI are collaborating on the workflow, with the AI orchestrating the steps.

What this means for you

Think about all the exceptions and problems that occur in manufacturing, distribution, customer service and other aspects of your organization - and the levels of bureaucracy and issue management that take place to keep everything running smoothly.

What “rent a human” hints at is a near future where a super intelligence can more seamlessly manage these issues, and the resources - such as people - to address them.

The question isn't whether this will happen in your industry - it’s more a matter of when. We believe the companies that start experimenting now, even in small ways, will have a massive head start as these systems mature. And based on the pace of the last six weeks alone, "mature" is coming faster than anyone expected.

What we’re building with this

At PreScouter, we've been deep in the weeds figuring out how to apply this technology to how companies actually operate (but focused on desk work - no human-renting required... yet).

The big idea we’ve had this week is to stop thinking of OpenClaw as a "personal assistant".

Out of the box, OpenClaw is designed as a personal assistant for one user. We need it to support 100+ users. This means not without mixing up all their conversations with OpenClaw or giving everyone equal access to sensitive systems.

Our reframe is to start thinking of OpenClaw as a caretaker - a system that tends to our business processes and systems of record.

For example, our team has to manage a lot of tedious metadata in the work we do: client names, deadlines, personnel assignments - it’s a lot of Salesforce data entry that is necessary, but eats up hours of our team's time every week. We’re working on setting up OpenClaw to handle this kind of administrative work, so our team can focus on the valuable work.

Want to see this in action?

In a few weeks, I'm putting together a live session to demo how all of this works - the agents, the workflows - the magic! If you want me to send you the invite, just reply with “DEMO”. (And a big thank you to everyone who's already signed up!)

Talk soon,

Dino

P.S.  If this felt like a lot, just remember: the fax machine freaked people out too. We'll get through this one together.