I spent fifteen years keeping other people's lights on. Now I'm building the lab that makes those lights invisible.

This isn't a note about AI. It's a note about giving real businesses the same operating leverage that infrastructure companies have had for a decade — and making it so quiet you forget it's there.

The pattern that wouldn't go away.

Across Canada, the United States, and India — from a network operations centre in Gandhinagar to a service-provider floor in Herndon to multi-site rollouts in Toronto — I kept seeing the same thing. Smart teams. Good products. And underneath, a thick stratum of repetitive, mission-critical work nobody had time to fix.

Tickets. Patches. Listings. Customer replies. The Tuesday follow-up to the Friday email. The thing that wasn't strategic but absolutely had to happen — and was eating the senior person who should have been doing something else.

For a decade I automated the network layer of that problem. Scripts. Monitoring. Predictive maintenance. AIOps before it had an acronym. But the people layer kept slipping back. There were always more jobs than there were humans willing to do them — especially the boring kind.

The moment something changed.

Late 2024, the same language I'd been writing into router configs and PowerShell runbooks became something an AI agent could actually read, reason about, and execute on. Not "summarise a PDF." Not "draft an email."

Take an inbound, qualify it, hand it to the right person with full context. Watch a marketplace listing for compliance drift and fix it before a vendor flags it. Triage a queue and clear the bottom sixty percent without anyone touching it.

The agents weren't smarter than my best engineers. They were patient. They didn't burn out. They didn't quit at two in the morning. And critically — they could be scoped, instrumented, and shipped like infrastructure.

That last word is what hooked me. Infrastructure. Not magic. Not a chatbot. Something a real operations team could deploy, audit, and retire when it stopped earning its keep. Something with a logged trail. Something a CISO could sign off on at the same speed they sign off on the firewall.

Why a lab, not a startup.

Most AI companies are startups racing for a feature. A few are platforms selling primitives. Almost nobody is operating the agents they ship.

I started eVamb as a lab because the work needs both kinds of people: those who design the systems, and those who carry the pager when one of them breaks at three in the morning on a Tuesday. The lab is the form factor of that conviction. Small team. Few products. Each one we operate. Each one designed for businesses that have outgrown manual but haven't grown into a platform engineering org.

"Lab" also means we get to be honest about what we don't know yet. We ship in increments. We retire what doesn't work. The pace is deliberate — not because we're slow, but because we'd rather ship four complete agents than twelve half-built ones.

What I believe about the next decade.

The future of business AI isn't a louder chatbot. It's quieter infrastructure — the kind you stop noticing, because it's just doing the work. The way nobody thinks about a router until it fails.

The agents that will matter ten years from now will be:

• Operator-grade — observable, audited, governed by the same discipline that runs the network.
• Specific — narrow problems solved completely will beat universal demos every single time.
• Quiet — the best automation makes the customer's experience feel like it had more humans in it, not fewer.
• Honest — clear about what they can and can't do, and where the gate stops them.

That's the bet at the lab. Not a thousand surfaces. Four product lines, each with a real operator's stake in the work. Each one shipped only when it can hold the pager at two in the morning without us in the loop.

What I'd ask you to remember.

AI is going to get loud over the next few years — in your inbox, on your timeline, in every pitch deck that crosses your desk. Most of it will be noise. Some of it will be very expensive noise.

The signal underneath all of it is simpler than the noise makes it look: real businesses need real work done, by systems that work the way infrastructure has always worked — quietly, reliably, accountably, and out of the way. That's the only thing the lab is trying to build.

If that's the kind of conversation you want to have — about the work, not the demo — you know where to find me.

← Back to the lab    Talk to the lab →