/now
What I'm focused on.
Updated February 2026
The Vision
Kubernetes for knowledge work
I burned $10K learning that AI agents don't scale like scripts. They scale like infrastructure. Now I'm building the control plane: agents, workflows, and human tasks as reconcilable resources.
The through-line
Everything I'm building points toward one goal: making AI agents as reliable as infrastructure. DevOps taught us how to make unreliable servers reliable. Same patterns, new domain.
Open Source
AgentOps
→A Go CLI + Claude Code skills plugin. The CLI (`ao`) compounds knowledge across sessions — forge, inject, ratchet. The skills run the workflow — /rpi, /crank, /vibe. Published and shipping.
Gas Town + gastown-operator
→Contributing to Steve Yegge's context orchestration ecosystem. Building the K8s operator that turns the CLI into a control plane. 27 PRs submitted, 8 merged.
Beads
→Git-backed issue tracking that survives agent sessions. 1,000+ issues closed through orchestrated workflows this month.
Kagent
→AI agent SDK for Kubernetes. Making agents first-class citizens in cloud-native infrastructure.
Writing
Day Job
Enterprise platform engineering
OpenShift clusters across air-gapped government networks. GPU infrastructure for ML workloads. The GitOps pipelines that make it scale without the team scaling.
Inspired by the /now movement. Updated when priorities shift.
Building in this space? I want to hear about it.