Cut the hype.
Ship real AI.

After SDN collapsed under its own hype, the networking industry was burned. The argument that followed was familiar: we can't trust software to run our infrastructure. Network engineers had spent years being told automation was the future, watched it fail, and were done being sold promises.

I spent the next several years helping change that narrative — not from the outside, but from inside a fast-moving Silicon Valley engineering organization at Apstra. I didn't just talk about what was possible. I built CloudLabs — a global cloud platform on AWS that let skeptical engineers and customers get their hands on working software instead of taking anyone's word for it. Let the product make the argument.

What made the difference wasn't just the technology. It was being thrown into a real software engineering team — not a network engineering team learning to code, but an actual startup engineering org moving fast. I had to learn to swim. I picked up the tools, the practices, the mindset. Then I went home on Saturdays and learned it the right way. The books, the materials, the foundations. Both things were necessary.

That background put me around the right people early. I was hearing about RAG and vector databases before they were products — before most people were even asking the right questions. I've been building with agents and LLMs long enough to know where the hype ends and where the real leverage begins.

The conversation I'm having with teams today about AI — the resistance, the skepticism, the "we tried that and it didn't work" — is the same conversation I was part of a decade ago with network automation. The difference is now I have the pattern recognition to explain why things are the way they are, not just what to do about them.