On June 1, GitLab cut about 350 people, roughly 14% of its workforce, and gave it a name: “Act 2,” a restructuring for the agentic era. The stated logic is that AI agents now handle code review, approvals, and the handoffs between them, so the company needs fewer humans standing in those loops and is reorganizing R&D into smaller teams with end-to-end ownership. It announced this in the same breath as a 23% jump in revenue.
That last detail is why this one is worth stopping on.
The Scapegoat Thesis, Tested
I have made the case that “AI did it” is mostly a cover story. It is the cleanest possible narrative for a CEO: not “we over-hired in the boom” or “we are protecting margin for the Street,” but “we are riding the future.” Most 2026 layoffs blamed on AI were, underneath, ordinary cost decisions wearing a fashionable jacket.
GitLab strains that thesis, on purpose. The usual scapegoat pattern needs a company under pressure looking for a flattering story. GitLab was not under pressure. It beat earnings. It is growing. And it still chose to cut and to put the agents front and center. When a winning company says the quiet part out loud, you have to take the claim more seriously than the reflexive cynicism allows.
Two Readings, One Costume
Hold both, because the truth is probably a blend:
- The honest reading. GitLab builds the DevOps pipeline. If anyone can credibly say agents absorbed the review-approve-handoff loop, it is the company that sells that loop. On this reading the work genuinely moved, and a smaller org is the rational shape of a business where machines do the connective tissue.
- The AI-washing reading. “Agentic era” is a gift to an earnings call. It reframes a headcount cut as visionary instead of corrective, flatters the stock, and spares anyone the admission that the 2022-2025 hiring was too aggressive. The agents are real; their causal role in the layoff is the convenient part.
— ByteIota, on the GitLab cutsAgentic Era or AI Washing?
The Tell Arrives Next Quarter
You do not have to guess which reading is right. The mechanism gives you a measurement.
If the eliminated roles really were absorbed by agents, output holds: GitLab ships at the same or higher velocity with fewer people, and the review-and-approval throughput that those humans provided shows up as automation in the numbers. If it was a costume, the work doesn’t disappear - it redistributes onto whoever’s left, velocity sags a quarter or two later, and the “agentic” framing quietly drops out of the next earnings call. The press release is a hypothesis. The next two quarters are the experiment.
What This Isn’t
- 14% is real pain. Whatever the framing, 350 people are carrying the cost of a narrative being tested in public. The analysis is cold; the layoff is not.
- Honest and convenient aren’t exclusive. The agentic story can be both genuinely true and extremely handy for the stock. Most corporate narratives that survive are the ones where the truth and the convenient align.
- One company is not the labor market. GitLab is a single, unusually well-positioned data point. It tells you the agentic layoff is now sayable on an earnings call, not that it is yet the norm.
The Quiet Part, Into a Microphone
For two years executives blamed AI for cuts while privately knowing the real driver was the balance sheet. GitLab did something rarer: it cut while winning, and it pointed at the agents on the record. Either it just gave us the first honest version of the agentic layoff, or it gave every CFO a better costume. We will know by the throughput, and so will they.
Related Posts
