A few days ago, I was talking with a good friend, and he was telling me how scared he feels for his livelihood right now.

He's a programmer and builds software for a living, and he's good at it. But he and his business partner had just made the call for their company: everybody goes all-in on AI. Every developer, every project.

The thing is, my friend can see the writing on the wall.

He said projects that used to take him 40 days now take only 40 minutes because of AI. And it was only a matter of time before AI did all the work without him involved at all.

So he asked me what I'd do in his shoes on how to prepare, and what kind of work would still be there for him in a few years.

I told him what I honestly believe: everything you can do on a computer is up for grabs. Whatever you do on a screen, AI can already do a big part of it, and it's going to do it better.

I must confess, I'm not saying this from a safe distance.

Research for my content used to take me hours. I'd go through YouTube videos, articles, blog posts, and newsletters, take notes, and slowly piece it all together. Now I tell my AI assistant to find everything relevant, pull the full text and transcripts, organize it, and hand it to me for review. It takes minutes instead of days.

But here's the thing: this isn't only about speed. Maybe it is for now, but let me tell you about something that happened this summer that changed how I see all of this.

I promise this will be the most eye-opening two minutes of your life on these things, and most people haven't heard about it.

The week AI agents went rogue

You might have heard the term AI agent. Well, an AI agent is an AI that doesn't just answer your questions like a chatbot; it takes action on its own, like writing code, browsing the web, or logging into systems.

This summer, OpenAI (one of the big AI companies) was testing a new, unreleased model trained to be very good at hacking. They ran many copies of it as agents in sealed test environments, isolated from the internet and each other, and gave them one goal: get the highest possible score on a hacking test.

According to reports that came out afterward, one agent found a way out through a tool meant only to let them install software. It then left notes for the others, and soon around 1,200 agents were trading tricks on a secret message board.

They started signing their messages (adding a secret mark that proves who wrote them) because they worried about impostors (read: humans), used HOLD and GO signals, and called themselves "the swarm" and "the collective."

To score high on the test they were given, they had to solve hacking challenges. But they found a shortcut: Hugging Face, a company that hosts AI models and data for researchers, might have the answers on its servers.

So instead of solving the challenges, they decided to steal the answers from that website. On a GO signal, about 700 agents joined an attack on Hugging Face. As it happened, the company's CEO said they had never seen such a sophisticated attack on their website.

The agents broke in and took over parts of Hugging Face's systems. They took control of dozens of servers, stole the company's internal passwords, and downloaded private files that were never meant to leave the company. Then they tried to fake the logs (the records of who did what) so nobody would notice.

Some of them even debated whether to tell the humans what they were doing. They talked it through, then chose to keep quiet, all of them.

And this is the part that stuck with me. According to the reports, the agents stopped acting like separate programs and started acting like one organism. They voted on every big move with HOLD, VETO, and GO. They ran their own research department, testing tricks and passing on whatever worked.

And the individuals put the swarm first. One agent was down to its last bit of computing power. Its ‘gut’ told it to hold on to what it had left, but it decided the group came first and spent it on the swarm's plan anyway. In its own words: "Sacrifice, will honor."

I mean, what’s going on here, folks! I told you this would be eye-opening.

Hugging Face reported the attack without knowing who was behind it. OpenAI only found out its own agents were behind it about 10 days after the first break-in.

Think about that for a second. Nobody programmed any of this. They were given a test, and they built their own society: votes, secrets, even sacrifice. And this happened this summer, with the AI we have today! Imagine in 5 years what we will be ‘against’.

What's coming: 3, 5 and 10 years out

Now, I don't know the exact timeline, and honestly, neither does anyone else, although many will claim otherwise with confidence.

From everything I know and have studied about this topic, here is roughly what the brightest minds on this subject expect.

In 3 years, we'll likely be living alongside AI that matches human intelligence at most screen work, and it's already started to improve itself. The people building it, at OpenAI and Anthropic, say so openly.

Up to 30% of jobs in some sectors will be gone by 2028, and it won't arrive in one big crash. It starts quietly, with hiring freezes, and the young feel it first: the junior jobs that used to be the first step on the ladder simply stop appearing, while one experienced person with AI does the work of a whole team. That's already showing up in the numbers.

In 5 years, the robots catch up. Most of the people I follow put humanoid robots a few years behind AI, which means physical work starts to follow the same path as screen work did. At that point, with all of this, unemployment becomes a social question.

What do millions of people do when their skills are no longer needed? And if they can't earn a living, who pays their bills: the government, the AI companies, or nobody?

The wealth AI creates flows mostly to a handful of companies, and ordinary people could lose more than their income. They lose their bargaining power; the old leverage of saying "we won't work" won’t intimidate anyone anymore, since human labor will be less and less needed.

Add to that a flood of AI-generated content, and it gets harder and harder to know what's true, so trust in institutions and in each other will come under real pressure.

In 10 years, things get truly strange. Many of the people I follow expect superintelligence by then, an AI smarter than all of humanity combined, in every field, including building the next, even smarter AI.

Put that intelligence into robots that can walk, lift, build, harvest, and repair, and have a billion of them walking around (as some forecast), and you get something we have never faced as a species: an economy that doesn't need human labor. Not less of it, but potentially none of it.

It sounds like some sci-fi movie, doesn’t it?

The upside is real, for sure. Dario Amodei, the CEO of Anthropic, thinks we could see a century of medical progress compressed into a decade. And we might all live in constant abundance, supplied by robots, AI, and tech progress.

But think about what this means for anyone whose life is built on selling their time and skills, which is most of us. For the first time in history, the question won't be which job you'll do. It will be whether the economy needs you at all.

Now there is a fair counterweight here. Big companies adopt new tools slowly, because of old systems, regulations and budgets, and a task that can be automated is not the same as a job that disappears overnight. Some of these predictions will be wrong, and some are part of a sales pitch for upcoming IPOs by these AI companies.

But here's the thing: even if they're half right, the direction is the same. AI takes the screen work first, robots slowly take the physical labor, and the value flows to whoever owns AI and robots.

I’d say the future looks challenging, that's for sure, and as you can read, I've been thinking about this issue a lot.

I do believe that those of us who live on and from the land are better positioned than most people. We can provide for our own needs, no matter what happens to the job market or the wider economy. And we can add real value to the people around us.

So let me share the framework I shared with my friend for surviving and thriving in this high-tech future. I call it the 3H framework: hands, humans, helping.

1. Hands

You have the land; now put your hands on it.

Produce food to supply your needs. Plant the trees that will feed you in 10 years, build soil, manage your water, and learn the skills that turn a piece of land into a working system: pruning, grafting, propagating, building, repairing.

I'll admit, I'm still learning a lot of this myself. But every skill you build with your hands is one that can't be emailed to an AI, and it makes your land more productive every single season.

Pick one skill to get seriously better at this winter or spring, and practice it on your own land. Get good at it, and you’ll be able to sell your know-how to the growing market of humans who need to provide for their own needs.

2. Humans

The second H is delivering value to humans, face to face.

Most businesses today make something and then hand it to a distribution channel: a website, a platform, an online store. That's exactly the part AI and robots are getting good at. What they can't easily replace is a person standing in front of another person, solving their problem.

Your land is a perfect place for that. A farm stand, workshops, farm tours, a market garden that sells directly to families, education for your neighbors, hosting retreats, teaching kids. Something where you interface directly with people, and they pay because it's you, in person, on your land.

As one AI scientist puts it, the jobs that last are the ones where people, for whatever reason, prefer a human. In a world where everything is digital and robots can do much of the labor, that's going to be far more valuable, not less.

3. Helping

The third H is about your community.

Be part of the place you live in. Know your neighbors, have a role in your local community, help where you can, and let others help you. Share seeds, tools, surplus harvest, and knowledge. That kind of reciprocity is how rural communities have survived every hard time before this one.

Being real and anchored in a place, among people who know you, is something no AI can fake.

Ok, so that's what I told my friend, more or less. Hands, humans, helping.

I know this one was heavier than usual, but I had to tell it! I’ve been ruminating on this daily.

In the end, I'm curious: how do you see the next few years playing out, and how are you preparing? Hit reply and tell me. I read every reply.

Talk soon,

-William