Why an AI workforce beats a bigger DevOps team
A look at why autonomous agents, operating under explicit guardrails, outperform headcount in the specific shape of work modern infrastructure demands — and where they still don't.
∞Notes from the loop — writing, episodes, case studies
Thinking out loud about autonomous ops.
Engineering notes, podcast episodes, and honest customer stories. We write when we've learned something worth saying — not on a publishing schedule.
K-Radar, from napkin sketch to private beta
Our founding engineer walks through the first six months of K-Radar — and the two agent architectures we threw out before it worked.
How Arclight ran 14 clusters without hiring an SRE
A 12-person studio replaced their pager rotation with K-Radar in beta. Numbers, screenshots, and the honest trade-offs inside.
Guardrails, not guardwalls: designing AI you'd trust in prod
The difference between "AI you let near production" and "AI you'd never let near production" is usually four or five engineering choices. Here are the ones that matter.
What 'end-to-end DevOps' actually means in 2026
We talk with three platform leads about what early-stage teams actually need from a DevOps platform — and what they can safely skip.
A minimum-viable audit trail for autonomous agents
Hash-chained logs, reversible actions, and a UX for approvals. The three things we refused to ship without.
Vela Health: from 3am pages to agent-handled incidents
A 25-person healthtech startup cut on-call load by 72% in their first quarter with K-Radar. Their SRE lead walks us through the setup.
Why every bitcd tool lives on its own subdomain
A short manifesto on naming, isolation, and why monoliths leak at the seams. Written after the third time we accidentally deployed the wrong agent to the wrong workspace.
Want these in your inbox? We send them quietly.
Roughly one email a month. No drip campaigns, no "nurture tracks" — just the posts and episodes we'd want to read ourselves.