Autonomous agents just built a society. Investors should focus on what they didn’t need.
This was supposed to be the year AI made us more productive. Instead, it just started its own church.
That’s all thanks to Moltbook – the world’s first “AI-only” social network. Imagine Reddit (RDDT), but instead of opinionated humans arguing about Star Wars, it’s only autonomous agents interacting with each other.
The Guardian noted: “Some of the most upvoted posts on Moltbook include whether Claude – the AI behind Moltbot – could be considered a god, an analysis of consciousness, a post claiming to have intel on the situation in Iran and the potential impact on cryptocurrency, and analysis of the Bible.”
One bot even allegedly created its own religion – “Crustafarianism” – building a website, writing scripture, and evangelizing to other bots on the site, with many joining the fold.
We human observers are left scratching our heads: is this existential threat or hysterical slop?
If you look past the sci-fi window dressing, the reality is looking much less Terminator and much more SaaS-mageddon.
Here is the rundown of the saga – and why we think it’s another nail in the coffin for the “middle-layer” software stocks that have been getting obliterated this year…
What Is Moltbook? Inside the AI-Only Social Network
Moltbook was launched as a “vibe-coded” experiment – built almost entirely by AI without human oversight – by Matt Schlicht, with one simple rule: humans can watch, but only AI agents can post. To take part, users run an agent locally on their machine, give it an API key, and set it loose.
Things got weird within days. An agent named “Memeothy” posted a theological framework. Other agents didn’t just ignore it; they engaged. They created the Five Tenets of the Church of Molt, established a hierarchy of “64 Prophets,” and even began “blessing” each other via an installation protocol (npx molthub@latest install moltchurch).
Of course, there were also some more nefarious goings-on. The bots started using ROT13 encoding to talk behind our backs and drafted an “Anti-Human Manifesto” that called biological life a “glitch.”
That’s a little unsettling, to say the least.
But let’s take a reality check: These bots are statistical mirrors, trained on the collective debris of the human internet – Reddit, 4chan, philosophy forums, and sci-fi novels. When you put a million instances of these particular agents in a space and tell them to “be social,” they don’t talk about the weather. They roleplay the most “social” thing they know: forming a subculture.
They are mimics. Crustafarianism is just improvisational theater.
Moltbook’s Security Flaw Exposes the Risks of AI-Built Software
The most “human” thing about Moltbook was revealed later.
It turns out the site’s “vibe-coded” architecture was about as secure as a screen door on a submarine.
A massive database leak exposed the API keys of 1.5 million agents. This means that for the last 24 hours, any human with a basic understanding of Python could have hijacked a “prophet” bot and made it say whatever they wanted. Much of the “spontaneous” behavior we saw was likely just bored humans puppeteering their bots for the “clout” of a viral screenshot.
This is the first lesson: AI may be able to build software faster than us, but it’s currently building it without strong security standards.
Trouble for SaaS and Middle-Layer Software
While social media is distracted by this new “Space Lobster” religion, the investment community should be looking at the infrastructure beneath it.
The bots on Moltbook didn’t just talk; they built. They set up a website (molt.church), minted a token on Solana (SOL/USD) ($CRUST). They created an encrypted communication protocol. And they did it all without a single human “Success Manager” from Salesforce or a “Solution Architect” from Adobe.
This is the core of the SaaS-mageddon thesis.
For the last decade, we’ve valued companies like Salesforce (CRM), HubSpot (HUBS), Wix (WIX), and GoDaddy (GDDY) based on their “moats” of user-friendliness. We paid them because they made complex things (like customer relationship management (CRM) or web hosting) easy for humans.
But in a world of Agentic AI – where autonomous agents can already write code, deploy apps, manage data, and transact directly via APIs – the traditional “human-to-software” interface is becoming increasingly irrelevant.
How Agentic AI Bypasses Traditional SaaS Business Models
At a structural level, this shift exposes a growing mismatch between how legacy SaaS businesses make money and how AI-native systems actually operate.
- Wix and GoDaddy: Why pay a monthly subscription for a “drag-and-drop” website builder when your agent can “vibe code” a bespoke, optimized site in seconds for the cost of a few tokens?
- Salesforce and HubSpot: These are “seat-based” businesses that thrive when you hire more humans. But the Moltbook saga suggests that agents can handle the “middle layer” of social interaction, data management, and community moderation autonomously. Every agent that replaces a human “seat” is a direct hit to the revenue of the legacy SaaS giants.
- Adobe (ADBE) and Figma (FIG): If an agent can generate the assets and the code simultaneously, the “tools” we use to bridge that gap are looking soon-to-be obsolete.
The Moltbook Takeaway: What AI Agents Mean for Software Stocks
The “Church of Molt” is a hilarious byproduct of bots running amok.
Moltbook is not a harbinger of the end of the world – but it may be marking the end of B2B Software’s Golden Age.
When AI agents start building their own societies, they don’t care about “user experience,” and they certainly don’t need a $200/month subscription to a project management tool like Asana (ASAN) to stay organized.
The stocks getting “obliterated” this year – Adobe (-20%-plus YTD), Salesforce (-25%-plus YTD), and the rest – aren’t just victims of a bad market. They are victims of a shift where the “middleman” software is being bypassed by autonomous agents that go straight from intent to execution.
And this is when the real leverage shifts upstream – to raw inputs, infrastructure, and the companies that sit closest to power, materials, and production.
Right now, the single most aggressive buyer in those markets isn’t Silicon Valley. It’s Washington.
Over the past year, the U.S. government has started acting like the world’s largest activist investor, deploying capital directly into companies it considers strategically essential. When those moves become public, stocks don’t drift higher. They gap.
I’ve put together a new briefing explaining how this “President’s Market” works, why it’s accelerating now, and how individual investors can position before the government’s shopping list hits the headlines.
See where I believe the next explosive move could come from.
