Blog

Practical notes on consulting, engineering, and operating modern IT.

Two industrial control panels on brushed metal, each with a large red mushroom emergency stop button and three key switches beside a turnstile.

Oct 19, 2026

The stop button nobody owns: Why AI agents need a kill switch before they need more autonomy

Somewhere in your stack an AI agent is still running a task nobody can cleanly halt. It opened a tool call, wrote a partial change, queued the next step, and kept a credential warm for the follow-up. Autonomy demos look impressive until you need to stop one agent, one bad run, or every agent that shares the same tool path. Then the room discovers the stop button was a slide, a chat hope, or a kill command only the person who wired the demo still remembers. Mid-size engineering and ops teams rar

Close-up of hands marking dates with a blue pen on a large paper desk calendar beside a keyboard, like a release schedule being planned.

Oct 16, 2026

Change freezes that never thaw: Why temporary stop-ships become your real release calendar

Somewhere in your release process sits a stop-ship that still sounds temporary. It started after a hard outage, a compliance scare, a board week, or a messy migration, and it was meant to last "a few days." Deployments paused, exceptions needed a verbal yes from whoever was around, and leadership relaxed because change risk finally went quiet. Weeks later the freeze is still the calendar. Nobody owns the thaw, the exit criteria were never written down, and the review date has slipped into folklo

Hand holding a blue pen pointing at a colorful stacked bar chart on a printed report.

Oct 15, 2026

Error budgets that never decide: Why SLOs without a decision owner become wallpaper

Error budgets without a living decision owner turn green SLO charts into wallpaper. Ownership of the burn decision beats dashboards alone.

Four colleagues gathered at a whiteboard with charts and papers in a bright modern office.

Oct 14, 2026

Postmortems that change nothing: Why blameless writeups fail without owned follow-ups

Blameless postmortems fail when follow-ups have no owner, date, or production evidence. Mid-size teams need a closure habit that turns learning into change.

Silver laptop and keyboard on a dark desk beside a small plant, lit by a bright desk lamp in a quiet office.

Oct 13, 2026

On-call without a charter: Why a pager without a contract burns the team

Paging without a living charter turns coverage into a name on a calendar. Treat on-call as an operating contract with owners, first-hour rights, and evidence.

Row of black server racks with dense red and blue cabling in a brightly lit data center aisle.

Oct 12, 2026

Config drift that nobody owns: Why production quietly stops matching what you think you run

Temporary console clicks and hotfixes quietly become unowned production config. Mid-size teams need named ownership of intended state, exception clocks, and a practiced reconcile habit.

Open blank notebook, stacked journals, coffee mug, pencil, and tablet on a dark wooden desk.

Oct 9, 2026

Runbooks that rot: Why the wiki fails the first real page

Peacetime wiki runbooks fail the first real page when nobody owns living docs. Treat recovery pages as operational surfaces with owners, verify dates, and proof under pressure.

Two white rocker light-switch panels mounted on a grey textured concrete wall.

Oct 8, 2026

Feature flags that never die: Why temporary toggles become permanent architecture

Temporary feature flags that outlive their launches quietly become permanent architecture. Ownership, consumer maps, and cleanup habits matter more than another flag dashboard.

Light-blue fiber patch cables with white connectors plugged into a vertical row of blue network ports.

Oct 7, 2026

Dependencies without owners: Why a vendor outage becomes your outage

Critical third-party and SaaS dependencies without named owners, detection, or a first-hour plan turn someone else's outage into yours.

Macro close-up of a brushed metal combination lock dial with engraved numbers.

Oct 6, 2026

Secrets that never rotate: Why long-lived credentials become permanent infrastructure

Static keys and service accounts that outlive their creators quietly become permanent infrastructure. Ownership, consumer maps, and rotation habits matter more than another vault.

Gloved hands turning a key in a brass padlock on a weathered red metal door.

Oct 5, 2026

Break-glass that never closes: Why emergency access becomes your real permission model

Break-glass access starts as a kindness to the future. Someone invents a fast path for the night when production is on fire and the usual approvals are asleep. A shared admin role, a standing exception, and a temporary vault that never quite expires. Then the outage ends, the glass stays open, and the exception quietly becomes how work gets done on ordinary Tuesdays. Mid-size teams invent emergency elevated access for real pain, then forget to treat closing it as part of the product. This post

Oct 2, 2026

Alerts without owners: Why dashboards alone never cut MTTR

Dashboards and alert floods do not cut MTTR when nobody owns the page, the runbook, or the next decision. Named ownership, actionable pages, and evidence that acknowledge means something do.