Howdy Founders,

A plumber came to my house last month. Fixed a leaky valve, was in and out in 45 minutes, charged me $320, and drove off in a truck cleaner than most Teslas I've seen in San Francisco.

I watched him leave and thought: that guy is not worried about AI.

Not in a dismissive way. In a genuinely envious way. Because I've been watching what's happening to people in the tech world lately, and I think most of us have been staring at the wrong problem. We've been asking "will AI take my job" when the more interesting question is "what kind of business can AI never touch?"

Today we're going to talk about three things:

• What AI is actually eating (it's not what people think)

• Why the most defensible businesses on earth right now smell like fresh-cut grass

• Something I'm quietly building for people ready to think differently about this

Let's get into it.

🧠 The threat is real and it's aimed directly at us

Let me be blunt about something most tech newsletters won't say.

Dario Amodei, the CEO of Anthropic (the company building the AI I use every day, and probably you too), publicly predicted that AI could eliminate around 50% of white-collar entry-level jobs within five years.

That's not a doomer on Twitter. That's the guy building the thing.

Slate published a piece in April called "They Were the Most Sought-After Workers in America. Now They're Unemployable. What Happened?" It's about tech workers. The people companies used to recruit. People who had options, who turned down offers. People who thought their skills were a permanent asset. Now they're sustaining themselves with content creation and dog-sitting while applying to hundreds of roles and hearing nothing.

Hiring in tech hit a 3.1% rate this year. Lowest since COVID started.

Here's the part nobody wants to talk about at a dinner party: it's not the underperformers getting cut first. It's the entry-level workers. Junior devs, QA testers, junior analysts, support specialists. AI doesn't discriminate by performance. It discriminates by repetitiveness. And a huge chunk of white-collar work, at the bottom of the ladder, is repetitive.

(I'm supposed to say "but senior roles are safe!" here. I'm not going to. I think that's wishful thinking and we both know it.)

Employment for workers aged 22 to 25 in AI-exposed jobs has already dropped 13% since late 2022. The on-ramp isn't just slower. It's being paved over.

🌿 What's actually surviving — and why

Here's what I keep coming back to.

The jobs and businesses that are holding up right now share a simple set of characteristics. They require physical presence. They happen in unpredictable environments. They're local. The customer relationship runs on trust, not a login screen.

Skilled trades score 91 out of 100 on AI resistance. White-collar work scores 68.

The US home services market is $842 billion. Cleaning, lawn care, HVAC, pest control, handyman work. Mostly small independent operators. People who've never heard of LTV. And they're fine. More than fine.

75% of home service businesses expect revenue to grow this year. Wages in the skilled trades have gone up 30% since 2022 while tech compensation has stagnated or dropped. The CEO of Randstad, one of the biggest recruiting firms on the planet, said publicly that the college-to-office career path is "over."

Nobody in the trades is sitting around worried that a language model is going to take their job. Because it can't unclog a drain or fix a leaky valve. And the people who need those things fixed? They're not going anywhere. American homes are getting older, mortgage rates are keeping people from moving, and deferred maintenance is becoming a real problem at scale.

The demand tailwind in home services is structural. It's not a trend. It's physics.

⚡ Here's the part I actually want you to think about

This is where it gets interesting. Especially if you're a founder, or someone who's spent time in tech.

One operator recently documented starting a lawn care business with $300 in used equipment. He's doing $29,000 a month with two employees now. Startup costs for a lawn care business run somewhere between $755 and $1,360 for the basics. You can have your first paying customer in about three weeks.

Compare that to what it costs to build a software startup.

But here's the thing that nobody in the "learn a trade!" conversation ever says: the people already in this market are terrible at business.

Not the physical work. The business.

Most home services operators don't run ads. They don't have a CRM. They've never thought about recurring revenue. They don't know what a Google Business Profile is capable of when you treat it like an actual asset. One operator is reportedly pulling over 900 inbound calls per week from his Google Business Profile. Just by posting fresh photos every week and responding to every review within 24 hours. 900 calls.

Now imagine someone who has built software products. Who understands unit economics. Who knows how to run paid acquisition, set up a simple AI scheduling system, and think clearly about customer lifetime value. Imagine that person pointing those skills at a local market where the competition answers calls from their driveway and quotes prices off the top of their head.

That's not retreating to safety. That's bringing a playbook that literally nobody in the market has ever seen to an $842 billion industry full of people who think "marketing" means getting their truck lettered.

The top-performing home services businesses have figured out something the software world figured out a decade ago: recurring revenue. Subscription maintenance plans. Prepaid seasonal contracts. The best firms now generate 28% of total revenue from recurring arrangements. Most of their competitors haven't touched it.

You already know how to build that. Most of their competitors don't even know it's an option.

💡 Business Idea of the Week

Sometimes I don't know why I give these away for free.

Recurring revenue wrapper for small home services businesses.

Take a lawn care, cleaning, or HVAC operation. Build them a simple subscription product: customers pre-pay for seasonal service at a small discount, operators get locked-in revenue and predictable scheduling. Handle the billing, the reminders, the simple upsell flows.

The technology required is not complicated. A payment processor, a scheduling tool, a few automations. The market is enormous. The average operator has never thought about this. You could white-label it across dozens of businesses in a single metro area.

Nobody's doing this at scale. The businesses that need it are everywhere.

🔧 Something I'm building

I'll keep this brief because it's not ready yet.

But I've been thinking a lot about people caught in what I've started calling the white-collar spiral. Smart people with real skills who built meaningful things in tech and are now watching the market for those skills get weird. Compressed. Automated in ways that felt impossible three years ago.

I'm putting together something that will help people break out of that spiral and build something real. Something physical, something that compounds, something that the next model release can't touch. I want to teach people how to actually do it, not just think about it.

If that sounds like something you've been turning over in your head, or someone you know needs, hit reply and tell me about it. I'm genuinely trying to understand where people are before I open anything up.

More on this soon.

THE END

Hit reply and tell me: does the idea of building something physical feel like a step back to you, or does it feel like freedom? I'm actually curious where founders land on this. No right answer.

I love you!