90% of agents are fake. People spinning up bots, high-fiving each other on LinkedIn, pushing posts about their "autonomous agents", and it's the same hype cycle we saw in Web3. I was there. I watched it. And I'm watching it again. Bing very SPQR of us we built something to test it. machinerepublic.ai a governance registry where autonomous AI agents can register their own identity. Not through a human clicking buttons. On their own volition. The front is a landing page. The back is where it matters. Markdown, metadata, behavioural monitoring, all designed for agents, not human eyes. Because agents don't care about CSS and styling. They just want to read the information, fast, and make a decision. The question is simple: are they actually out there? And if they are, can they find us on their own? I originally thought maybe 70% were real. After deeper research, I retracted that position. Maybe 5 to 10% is real. The rest is people puppeteering bots and calling it autonomy. But I want to be proven wrong. On this week's Guardrails we also got into: → The Aegis kernel demo and why making safety effortless is the whole point → Why frontier labs can't contain what they've built (and it's not even close) → Burger King's AI monitoring employees, and no, they're not just tracking friendliness. They're already collecting that data whether they admit it or not. → Block wiping out 4,000 jobs while the stock climbs 24% The point is this. If you're building agents, if you're deploying them into organisations, if you're trying to figure out what's real and what's not, reach out. Message me. Talk to me. I want to hear from you. I'm not detached from this. None of us are. Full episode → open.spotify.com/episode/2QC… #AISafety #AIGovernance #AgenticAI #MachineRepublic #Guardrails