Originally published at devopsdiary.blog. Series opener for "The Quiet Years," a retrospective on the work between August 2022 and now.
The last post on this blog went up in August 2022. Three and a half years later, here's why the silence happened and why it's ending now.
April 14, 2026 · Todd
One of the last post on this blog went up on August 2022. Time to restore service. Twenty-eight articles in nine months, and then nothing for three and a half years.
I wasn’t burned out. I didn’t lose interest. The blog went quiet because I took a new job two weeks later, and the work ate the writing.
That’s the honest version. The strategic version, the one that matters now, is that the work itself was the foundation I needed for what I’m doing today. I just couldn’t see that while I was inside it.
The work that ate the blog
In August 2022 I started a new Technical Architect role. I thought I’d be enabling DevOps practice. What I ended up doing was a lot of the day-to-day firefighting that comes with a large enterprise.
I spent the better part of 2023 in conference calls explaining why declarative deployments didn’t violate change management.
While that was happening, I was also running vendor evaluations and designing the configuration automation for our public cloud alongside an existing CloudBees installation. I built dashboards nobody wanted to see and I figured out what to do when Anaconda changed their licensing and hundreds of developers were impacted. A dev container solution I prototyped for my own team ended up getting adopted.
None of that looked like blog material at the time. It felt like work. It was the daily grind of making enterprise engineering slightly less terrible one approval at a time.
What I didn’t see coming
Somewhere in the middle of that stretch, ChatGPT showed up. Then Copilot. Then a flood of other tools that could generate code faster than any human could review it.
My first reaction was skeptical. My second reaction, was something like “how the hell are we going to govern this?” The output wasn’t bad. It was often impressive. But it was also a black box. It was a new source of engineering artifacts that could be produced at scale, but with no clear way to validate them or trace them back to the decisions that led to them. Architecture documents, design specifications, PRDs, test cases, deployment scripts. All of it could be generated by AI, but none of it could be governed by the processes that had been in place for human-generated artifacts.
That observation is where the rest of my career bent.
The governance instincts I’d been building (immutable artifacts, structured handoffs, validation that produces verdicts instead of suggestions, measurement that becomes gating) turned out to be the vocabulary AI-assisted software delivery needed. And almost nobody was connecting those dots. The MLOps world was building model training pipelines. The AI safety world was talking about alignment. The engineering leadership world was dreaming about productivity gains.
The gap in the middle was empty. Nobody was writing about what governance looks like when AI generates engineering artifacts at scale. That gap is where I’ve been living since early 2026.
Why now
In February I started building AIEOS, an open-source governance system for AI-assisted software delivery. I wrote the first post about it two weeks ago. That post is the reason this one exists.
I can’t keep writing forward-looking pieces about AI governance without also explaining where the ideas came from. They didn’t show up in February. They came from watching engineers try to absorb new tooling while keeping regulatory commitments, audit trails and production reliability intact. That’s the blog I didn’t write while I was living it.
So I’m going to write it now, in retrospect. This retrospective won’t read like greatest hits. Several of these posts are about things that didn’t work. A couple are about decisions I’d make differently today. I’m not trying to stack up wins. I want to show the actual path from doing enterprise governance to building AI governance infrastructure, because that path is shorter than most people think, and a lot of engineers are walking it right now without realizing it.
If you’re one of them, this series is for you.
Todd Linnertz is the creator of AIEOS, an open-source AI governance system for software delivery teams. Find him at devopsdiary.blog and github.com/wtlinnertz.
Top comments (0)