The Day I Stopped Checking My Pull Requests
It was a Thursday in March 2024, and I had seventeen open pull requests sitting in the queue. I know it was seventeen because I counted them later, the way you inventory a house after a flood — not to fix anything, just to understand the shape of what you lost. I had been a software engineer for fifteen years. I had shipped features at two startups and one mid-size company, survived three rounds of layoffs that took colleagues I admired, and built enough institutional knowledge that my manager used to joke I was "the last person you'd ever want to lose." The joke stopped being funny when she had to tell me the decision wasn't hers.
The specific moment I want to describe isn't the conversation where I was let go. That part is blurry in the way that bad news always is, the words arriving slightly ahead of your ability to process them. The moment that actually changed something in me came about two weeks later, when I sat down at my laptop out of pure muscle memory and opened GitHub.
I had no repositories to push to. No JIRA tickets. No standup in forty minutes. I just sat there looking at a green contribution graph that stopped abruptly, like a city map that runs out of roads. And I felt something I hadn't expected: not grief exactly, but a strange vertigo at having organized so much of my identity around a set of daily actions that could simply be — switched off.
I had been good at my job. That's not vanity; it's the detail that matters most to this story. I wasn't displaced because I'd stopped learning or grown complacent. I had been learning constantly, in fact — picking up new frameworks, running internal workshops on tooling, mentoring junior engineers. But the company had deployed a code-generation suite that, combined with a leaner team, could cover what four mid-to-senior engineers had been doing. The economics were not subtle. Goldman Sachs had been writing about this dynamic for two years by the time it reached me; the Oxford Martin School had been writing about it for longer. I had read those reports with the mild interest of someone watching weather develop over a distant mountain range.
The mountain came to me.
What I learned in the weeks after I stopped checking those pull requests was not that I had been naive — though I had been — but that my mental model of risk was structurally wrong. I had been calculating my job security the way most people do: comparing myself to my peers, tracking my performance reviews, keeping my skills current relative to other humans in my field. What I had not done was track how my role was being repriced relative to what a model plus a smaller team could do. Those are different measurements entirely, and I had been taking only one of them.
There's a particular loneliness in being technically sophisticated and still getting caught flat-footed. Engineers are supposed to see systems clearly. We understand automation as an abstract principle; we have built automating tools ourselves. But understanding something as a concept and understanding how it applies to your own cost-on-a-spreadsheet existence are separated by a gap that is apparently quite easy to fall into. I know this because I fell into it, and because since I started writing publicly about displacement, hundreds of people in technical roles have written to tell me they fell into it too, usually using phrases like "I thought I was safe" or "I never thought it would be this fast."
The speed is the part that keeps striking me. Not dramatic, not cinematic — just quiet and administrative. A meeting, a package, a laptop return label. The pull requests sat unreviewed for a few days and then were reassigned. The work continued without a gap you'd notice from the outside.
I'm not telling this story to produce despair. I'm telling it because I think the specific texture of that moment — sitting at a laptop, logged out of a professional identity I'd held for fifteen years, feeling not dramatic tragedy but a kind of baffled arithmetic — is more useful than any generalized warning. The warnings are everywhere. What's rarer is the granular, embarrassing, specific account of what it actually feels like to have been a capable person who measured the wrong things.
I started measuring different things after that. I looked at which roles inside my former company had grown versus shrunk over the previous three years. I looked at where automation investment was flowing, using earnings calls and SEC filings as source material rather than tech journalism. I looked at my own skill set not as a credential but as a portfolio with exposures. It took months of that work before I understood the landscape well enough to feel steady in it again — and that process of understanding is what eventually became Clankers, the displacement-readiness project I now run.
The seventeen pull requests were never merged. I checked once, out of curiosity, about six months later. They had been closed.
PROTECT YOUR INCOME NOW
If AI automation threatens your job, preparation is the only honest answer: know your risk, build a runway, and reskill before you have to. The Clankers 90-day programme structures exactly that — free, which is exactly why today is the right time to start.
Start the 90-Day Programme — Free →GET THE CLANKERS BRIEF — FREE
One email every Monday: the week's AI displacement moves that matter, and what to do about them. No spam. Unsubscribe anytime.