The Missing Infrastructure Nobody Is Building for Zero-Human Companies

Here's a number that should recalibrate your week: 17 specialized agents for less than one month of a junior developer's salary.

That's the economic argument being made right now around zero-human company orchestration, and it's hard to ignore. A solo founder named Aleister is reportedly doing $11.9K in monthly revenue at zero headcount. Pre-configured agent orgs are being packaged as downloadable products. The pitch is basically: stop hiring, start downloading.

But before you go all-in on that pitch, it's worth slowing down and looking at what's actually missing from the stack underneath it.

The coordination framework getting the most attention right now hit 53,000 stars in six weeks. That growth is real. The hype-to-substance ratio, though, is a legitimate concern that even the honest breakdowns of this space are calling out. Fast-growing repos aren't always the most architecturally interesting, and in this case, some of the more thoughtful engineering is happening in projects with far fewer stars.

One example worth knowing about: there's an institutional veto mechanism in a project called Edict that borrows structure from the Tang Dynasty's administrative system. Sounds esoteric, but the problem it solves is real. When you have cascading agent delegation, you need a layer that can stop a decision before it propagates. That's not a feature the dominant framework even discusses. Another project called ClawCompany has explicit model cost routing built in, meaning it actively chooses which model handles which task based on price. Neither of these patterns is getting much airtime, but both solve problems you'll eventually hit in production.

Speaking of production, there are two engineering details worth understanding if you're evaluating any of these tools for real work. The first is atomic task checkout, which prevents two agents from grabbing the same task simultaneously at the database level. That's not a nice-to-have, that's the difference between a system that works and one that doubles your spend and corrupts your outputs. The second is hard budget limits that stop agents mid-task rather than completing the work and sending you a bill afterward. If you're running anything beyond a weekend experiment, those two primitives matter more than almost anything else in the feature list.

Now here's where it gets structurally interesting, and honestly a little uncomfortable. The coordination layer handles what happens inside the company. But nobody is really shipping the infrastructure layer beneath that.

Specifically: how do agents find each other across systems? Right now there's no real agent discovery registry. There's no standardized trust scoring for agents operating across organizational boundaries. There's no verified identity layer that lets one agent confirm another agent is who it says it is. And there are no encrypted agent-to-agent communication channels built into these stacks by default. These are the gaps that will matter enormously once zero-human companies start interacting with each other or with external services at scale.

Think about what that means operationally. You're routing decisions through agents. Those agents are calling other agents. And the identity and trust infrastructure to verify any of that is mostly absent.

Which brings up the liability question, and this one has no clean answer yet. When an agent CEO approves a contract five delegation layers from the human chairman, current legal frameworks have nothing to say about that. The accountability chain is genuinely broken. Every zero-human company repo quietly sidesteps this, but it's not going away. At some point, one of these agent-approved decisions is going to land in a courtroom, and we're going to find out very quickly that nobody thought through who owns the output.

So here's how I'd synthesize all of this. The economics of zero-human companies are real and they're moving fast. The coordination frameworks are functional enough to build on. But you're building on a stack that has meaningful gaps in discovery, trust, identity, and legal accountability. The founders who take that seriously right now, before it becomes a crisis, are the ones who'll have defensible operations when this wave matures.

The $27 skill pack is a compelling headline. The infrastructure below it is where the actual work is.