AI Is Not a Labor Crisis...

Sam Lessin

Context:
Historically, we’ve always needed a story… a reason to get out of bed, to strive, to suffer, to build. Meaning isn’t a luxury. It’s the critical input to civilization. Because we can imagine the future, we literally can’t live without a reason to inhabit it.

We’ve weathered labor shocks before. We can probably fend off killer robots, and maybe even survive the collapse of shared language. But if we kill meaning, we kill civilization.

Market Signal:
The cracks are already visible. The old “industrial” meaning story:work hard, move up, make life better for your kids, has been breaking down for decades. First, the Mike Mulligan problem: physical labor stopped being enough. Then the college credential treadmill broke. The internet destroyed local hero systems and made all status relative, global, and zero-sum. Now, AI comes along and threatens to snap the last remaining ladders.

Meaning is becoming the real scarcity. Not intelligence, not productivity, not even stuff. But trust, belonging, purpose, and the sense that your striving matters.

Takeaways:

Yoni Rechtman

Context:
If you’ve been anywhere near Software Twitter, you’ve seen the meme war raging: “It’s all about SBC!” vs. “It’s all about terminal value!” (For the uninitiated: SBC = stock-based comp, TV = terminal value, i.e. what’s the business worth at the end of the rainbow.) The answer, as usual, is “both” but the real question is why both matter, and what that says about where the SoftWars™ are headed.

Market Signal:
Here’s the harsh truth: the market has decided that >95% of software is a finite annuity with $0 terminal value. If you’re running a business that’s basically an income stream, then you absolutely have to care about income and constrain costs (SBC included). If you’re one of the blessed few that the market sees as a potential perpetuity (a business with real terminal value, a compounder) then the rules are different. You can “overspend” on SBC because you’re buying acceleration, not protecting margin.

The dichotomy is stark: most SaaS is in the former bucket. Cutting SBC or running a tighter ship might make you more attractive to a PE buyer, but it doesn’t give you a future. The market is telling us: middling software businesses, even if run better, are still weighed down by the narrative albatross unless “run better” means “go all-in on AI and acceleration.”

Takeaways:
So, how does software fight back? You build. Not the slow, deliberate, multi-year roadmap kind of build. The world is moving too fast for that. You need to carve off a piece of free cash flow and empower a labs group (real mandate, CEO support, no legacy baggage) to invent new stuff for your customers. The only defense is a good offense, and in this AI era, prevent defense is a death sentence.

This is less about process tweaks and more about rewiring the company’s DNA: hire and promote the ones who get it, move on from those who don’t. Compensation and talent assessment will have to change.

And here’s where it gets personal: this matters at seed more than anywhere. The only questions that matter early are: Do I love this founder? Is there a story that matters? Is there something convex/interesting to spend money on today? Is there terminal value?

Forget “what’s your moat”, you can’t have one early, and if you did, you probably couldn’t have started at all (unless you’re the rare regulatory capture play). The point is to have a thesis about long-term margins, earning power, and how defensibility might emerge. If you don’t have a POV on terminal value 10 years out, you shouldn’t be investing in categories with 10+ year hold periods. This is true for founders and investors alike.

I find myself saying more and more: “You can do X, make some money, but if there’s no TV, what’s the point?” It’s not academic, it drives what you build and where you allocate capital, right now.

Sam Lessin

Context:
Lately, I’ve been mulling over what happens to the classic software “development shop” (Think Pivotal Labs, ThoughtWorks, etc.) in a world where building software is getting easier by the day. Do these shops go away? Or do they actually get bigger? Here’s my personal lens on where things are heading:

At this point, I’m cranking out one or two fully functioning pieces of software a week. Full-on apps and services, built exactly to my own spec, that I actually use in production to get things done. These aren’t just toys… they’re real, useful, and in theory, could be used by others too.

But here’s the rub: I have zero interest in the grind of fully productionalizing them": hardening for scale, scrubbing for weird security edge cases, worrying about uptime, or, God forbid, supporting users. I’d happily pay someone to do all that for me: take on the liability, manage the app, maybe even market and grow it. But I don’t want to product manage a dev shop, or iterate with them on features… I want to ship my working code, hand it over, and walk away.

Market Signal:
This is where the new “shop” comes in. Not a dev shop in the old sense, but a productionalization and management shop. I don’t want them to build software for me. I want them to take my already-built thing, make it bulletproof, support it, and maybe even commercialize it. I’d pay for that, and I suspect I’m not alone.

“But Sam, isn’t all software dead? Why wouldn’t people just build their own?” Fair pushback. I do think that’s the broad arc of history. But there’s a middle ground here. I’ve got at least 3-4 projects I’d pay someone to productionalize and manage—and maybe even make money from. I did this last summer with a newsletter aggregator: built it, handed off the code, said “make this for real.” It worked, but it wasn’t seamless.

And it’s not just me: at our recent creator-bot hackathon, people built amazing things with Claude—stuff they need and use—but they’re not going to cowboy it forever. They’d love to pay someone to support and run their creations.

Takeaways: