landscape · Julien de Waal · 11/14/2026 · 5 min read
One Developer vs. the Bot Internet: How Solo Founders Are Solving Human Verification Without Big Tech
The internet has a bot problem no one wants to admit
Estimates vary, but somewhere between 40% and 65% of internet traffic is non-human. Bots scraping, bots clicking, bots filling forms, bots pretending to be users. The infrastructure built to stop them—CAPTCHAs, behavioral fingerprinting, cloud-based biometrics—is owned almost entirely by a handful of companies: Google, Cloudflare, a few VC-backed challengers.
And now that same infrastructure is being quietly undermined by the very AI labs that depend on it.
Sebastian Kläy, a solo founder based in Bern, Switzerland, laid this out plainly in a post on dev.to. The core argument: every major AI provider is making the bot problem worse, while simultaneously offering tools—like MCP (Model Context Protocol)—that let agents and web services query through a standard interface. The result is a web where proving you're human is getting harder, and the people building the solutions are the same people who created the problem.
Kläy's response was to build OpenOS / EDNA: a human verification system that doesn't route your identity through a cloud provider.
Why big AI labs can't build this
The technical barrier to building decentralized human verification isn't insurmountable. The economic one is.
Cloud biometrics is a business model. When Google or Apple verify your identity, they're not doing it out of altruism—they're accumulating behavioral data, training signals, and platform lock-in. A verification system that doesn't route through their infrastructure is a direct threat to that model.
So the big labs face a structural conflict: they benefit from the bot economy (more queries, more API calls, more usage data), and they benefit from owning verification. Building a genuinely open, privacy-preserving alternative would cannibalize both.
This is exactly the kind of gap where solo founders operate. No board to answer to. No existing revenue line to protect. No partnership agreements that make a competitor's problem your political problem.
Kläy's thesis, stated directly in his post: the big AI labs can't build this—not technically, but economically.
What EDNA actually does
OpenOS / EDNA is building human verification infrastructure designed to work without centralized biometric databases. The specifics of the technical implementation aren't fully public, but the architecture philosophy is clear: verify humanity at the edge, without a cloud intermediary holding the receipt.
This matters for several reasons:
- Privacy: Centralized biometric databases are single points of failure. Breaches at identity providers have exposed millions of users. Edge verification reduces that surface area.
- Agent compatibility: As AI agents become autonomous company infrastructure, they need to interact with services that can distinguish human-initiated requests from agent-initiated ones—not to block agents, but to route them correctly.
- Economic independence: Services that rely on Google's reCAPTCHA or Apple's device attestation are one policy change away from a pricing or access problem. Decentralized verification doesn't have that single point of commercial failure.
The MCP angle is particularly sharp. As more AI systems adopt the Model Context Protocol as a standard interface for agent-to-service communication, the question of *who is making this request—a human, an agent, or a bot pretending to be either*—becomes foundational infrastructure, not a nice-to-have.
The solo founder advantage in infrastructure plays
There's a counterintuitive pattern in AI-native company formation right now: the infrastructure problems that look too boring or too political for big companies are exactly where solo founders are finding durable ground.
Human verification is a perfect example. It's not glamorous. It doesn't generate press releases. But every SaaS product, every marketplace, every agent-powered service needs it. And the current solutions all come with strings attached.
Kläy isn't the only one operating in this space with this logic. The same structural argument applies to payments infrastructure, compliance tooling, and data provenance—categories where the dominant players have conflicts of interest baked into their business models.
What makes the solo founder position viable here isn't just speed. It's alignment. A solo founder building privacy-preserving verification has no incentive to harvest the data. That's not a marketing claim—it's a structural fact. No investors demanding a data monetization strategy. No partnership team negotiating with the companies whose data moat you're disrupting.
This is the core of the one-person-unicorn model: pick a category where incumbents are economically prevented from solving the problem correctly, build the right solution, and let the conflict of interest do your competitive work for you.
Revenue per employee as a forcing function
One underappreciated consequence of building solo in infrastructure is what it does to your unit economics. Kläy doesn't have a 40-person team. He doesn't have a sales org. If EDNA gets traction, the revenue per employee ratio will be extraordinary by traditional startup standards—because there's one employee.
This isn't just a vanity metric. High revenue per employee in infrastructure means:
- Pricing power: You can charge less than a staffed competitor and still have better margins
- Survival: You don't need a Series A to keep the lights on
- Speed: Decisions happen in hours, not sprint cycles
The companies that will own the next generation of internet infrastructure won't necessarily be the ones with the most engineers. They'll be the ones who built the right abstraction early, with the right incentive structure, and stayed alive long enough for the market to catch up.
What this signals for the agent economy
The bot verification problem is about to get dramatically more complex. Right now, bots are mostly scrapers and spammers. Within two years, they'll be autonomous agents running marketing campaigns, negotiating contracts, and making purchasing decisions on behalf of real businesses.
The infrastructure question isn't just "is this a human?"—it's "is this a human, an authorized agent acting on behalf of a human, or an unauthorized bot pretending to be one of the above?"
That's a three-way verification problem. Current CAPTCHA infrastructure wasn't designed for it. Cloud biometrics providers don't have an economic incentive to solve it cleanly. And MCP, for all its elegance as a protocol, doesn't include identity attestation by default.
Kläy saw this gap. He's building into it. That's the job.
Sebastian Kläy's work on OpenOS / EDNA is one of the clearest examples of a solo founder identifying a structural market failure—where incumbents are economically blocked from the correct solution—and building directly into that gap. Watch the space.
---
Is your company eligible? Submit to the leaderboard → onepersonunicorn.co/submit
Read the full AI-native companies guide.
Is your company eligible? Submit to the leaderboard →
Submit Your Company