Welcome to Slow Ventures’ Snailmail, where we slow down and share what’s been on our partners’ and founders’ minds this week.
TL;DR:
OpenClaw Is Anti-OpenAI (And That's Why It Matters)
Sam Lessin
Context:
We just got back from Jackson Hole. Wheeling and dealing, epic powder, and hosting Jess and I’s annual mini-conference with founders, investors, and media folks.
Here's what I'm seeing: The founders are working really hard. Like, 24/7 hard. It's truly relentless right now. We had founders staying in our houses, and they're just going, you know? The work cycle is insane. You go to bed, you wake up, and the world has changed because there's so much energy in the system. So many people lit up. Even semi-technical people are contributing in interesting ways with AI now.
Market Signal:
OpenClaw is not good for OpenAI.
OpenAI's whole strategy has been monolithic. You write to their centralized service. They're the God machine. The early branding was all about AGI—the race to be first, winner-take-all. Very centralized.
OpenClaw is the opposite. It's the decentralized version. And here's why that matters:
The heavy brute force of LLMs is not going to be winner-take-all. It's not a race to AGI. It's going to be a bunch of backends that can do it—like multiple cloud providers. The interesting game becomes on the edge.
If the future is decentralized—people running daemons (background processes) on their own machines, all interacting with each other in a distributed cloud—then OpenAI's vertical integration doesn't work. Yes, they can still sell tokens. But that's bare metal. Low margin. Fair market competition.
For OpenAI to be worth what it is, they need this story: We're so much better that we subsume all the noise. We're the top of wallet. We're your consumer experience. We're the advertising interface. But if activity moves to open source and distributed edge computing? That story falls apart.
The real question is trust and sovereignty. There's a dystopian world where you give up everything to ChatGPT—every password, every piece of data, full sovereignty. That's valuable for them, but it's not happening. For two reasons:
You won't trust them with it.
They won't want the liability.
Right now, if you asked someone, "Do you want to give all your stuff to ChatGPT?" they'd say no. And if you asked ChatGPT, "Do you want to be responsible for everything a person does with you?" they'd also say no. Because some people will do malicious or semi-malicious things, and you don't know what's going to play out.
The decentralized sovereignty model moves faster. People will be way more experimental and do weirder stuff because no central entity has all the liability. You run your own computer in the cloud or locally. You give it what you want. Your responsibility. That's the world that wins.
The irony of all this? It plays to Apple's hands in a weird way.
I'm a perennial Apple hater, but I love how little they're spending on CapEx. Everyone else is spending hundreds of billions of dollars. Apple spent a few billion—less than they did last year. They literally went down on AI CapEx.
Takeaways:
OpenClaw is anti-OpenAI. Decentralized, personal AI sovereignty > centralized monolith. The interesting game is on the edge now, not the foundation models.
Apple wins by doing nothing. They're spending less on AI CapEx while everyone else burns billions. Mac minis become the personal AI hardware of choice.
The definition of "developer" is collapsing. Millions of new people are writing software for themselves in English. That's a good thing.
Sovereignty > centralization. You can't have a monolith take all the liability. Decentralized systems move faster because everyone owns their own risk.
We're back to 2007 energy. Hacker meetups, tinkering, showing what you've built. The grind-core SaaS culture of the 2010s gave way to this new wave of personal AI exploration.
The Death of SaaS Margins: Why Pure Software Is Now a Services Business
Yoni Rechtman
Context:
Let’s talk about the elephant in the room: “Software is dead.” Except, it’s not. Not even close. But the nature of software businesses—especially SaaS—has fundamentally changed. We’re living through a shift where the classic SaaS playbook doesn’t work like it used to, and the underlying economics are getting dented, if not outright broken.
Market Signal:
The signals are everywhere. SaaS TAMs are flattening, pricing power is eroding, and the once-magic 80-90% gross margins are now a myth for most. AI is the culprit: it increases marginal costs (thanks, inference!), accelerates value depreciation (tools improve at warp speed), and destroys switching costs (when everything is easy to replicate and agents don’t care what software they use, why stick around?).
Margins that used to be 25% are now 10-11%. Pure application software is starting to look a lot like the services business: easy to start, rarely scales big, and doesn’t need much outside capital. The old assumptions—perpetual growth, stable high profits, defensible moats—are going away for most.
But don’t get too bearish. TAMs aren’t going to zero, and the public market isn’t going to wipe out software entirely. No serious company is going to rebuild everything in-house, or trust upstarts with mission-critical workflows. Leaders still want a “throat to choke.” They want defensible vendor choices, best practices, and to avoid undifferentiated cost centers. Capital allocation is still king: focus on what you do best, outsource the rest.
Takeaways:
Software isn’t dead, but SaaS as we know it is. The old SaaS model—high margins, high multiples, perpetual growth—isn’t the default anymore.
Switching costs are the new moat. It’s not about code or product features. It’s about network effects, multiplayer dynamics, and embedding yourself so deeply that leaving is more painful than staying.
Sin-eaters win. If you can absorb the hairy, high-stakes, high-liability parts of a business, you create real value. The code is just the interface—what matters is taking on risk and complexity others want off their plate.
Hybrid models are the future. Pure software is too easy to copy. Bundle it with services (forward-deployed teams), hardware (proprietary data, physical lock-in), or vertical integration. Decommodify code by making it just one part of a bigger, stickier solution.
Brand, trust, regulatory capture, and distribution matter—but you can’t count on them from day one. Build for network effects and switching costs first.
The biggest opportunity ahead? Connecting foundation models to real-world, high-value use cases—through new, hybrid business models, not by rerunning the cloud/SaaS playbook. The future is brighter for software as a tool, not as a standalone business model. Let’s build accordingly.
Five Niche Creators, Through Our Lens
Megan Lightcap
Context:
Longtime readers might recall that back in May 2023 (!!!), I wrote about how we identify attractive niches and verticals. The tl;dr is that our approach is less about ticking off boxes or chasing hot categories, and more about digging deep into consumer needs, behaviors, and psychographics. In other words: we’re not just looking for “the next big thing,” but for creators who have a real, nuanced connection with their audience.
Market Signal:
This thematic sourcing and evaluation borrows from traditional consumer investing. We’re always asking: why does this creator matter to their audience? Why does a fan seek them out, follow them, and invest their time (and often money) in them? The market is telling us that deep, authentic engagement – not just raw reach – is what drives true significance and staying power.
Surprisingly (or maybe not), the core “categories” we identified almost three years ago are still guiding our interests today. That’s a signal in itself: consumer needs and the fundamental drivers of engagement don’t shift overnight, even as platforms and formats evolve.
Takeaways:
Let’s revisit the five frameworks that continue to shape how we spot opportunity:
Importance of Hobbies: In a world that feels more stressful by the day, people are turning to skill-based hobbies for relief, creativity, and a sense of accomplishment. Creators who serve these communities aren’t just entertainers – they’re building deep, trust-based relationships with audiences who have a real willingness to spend on their passions.
Recent example: @whatsthestitchndlpt – Emma’s needlepoint community is a modern, hands-on antidote to screen fatigue.Widening Knowledge Gap: As life gets more complex, consumers are desperate for guidance in areas where the stakes are high and the information is overwhelming. The best creators here are true experts, distilling complexity into actionable advice.
Recent example: @secretCFO – real-world, billion-dollar CFO insights for finance professionals.Taboo Topics: Some conversations are just too sensitive for the group chat. Creators who can build safe, judgement-free spaces for these topics foster intense loyalty and trust – and, over time, higher lifetime value.
Recent example: @seggstalkradio – Thea’s honest, open conversations about sexuality.Paradox of Choice: Too much choice is paralyzing. Enter the tastemakers, curators, and reviewers who help audiences cut through the noise. In nuanced, subjective categories, trust is everything.
Recent example: @gocleanco – Sarah’s no-nonsense home cleaning tips have built a cleaning army.Cultural Milieu: Social media has democratized representation. Creators who authentically share their lived experience – whether that’s a specific place, culture, or identity – are building new forms of community and resonance.
Recent example: @townsends – Jon’s immersive 18th-century content is a masterclass in time travel (seriously, if you know him, intro me!).
Asks:
If you are, or know of, a creator who fits within any of these frameworks, I want to hear from you. We’re always looking to connect with people who are building meaningful, durable communities in these spaces.
More Musing From The Team