GitLab restructures around the agent thesis, and Simon Willison checks the incentive

Simon Willison's post is short but the analysis earns its place. GitLab is reorganizing for what it calls Act 2: pulling back from small-team operations in up to 30 percent of countries, going from nearly 60 to roughly 42, flattening management by up to three layers, and splitting R&D into about 60 smaller empowered teams, nearly double the independent count. It also retires its CREDIT values for a shorter set built around speed with quality, ownership, and customer outcomes.

Willison's contribution is not the summary but the skeptical read. He finds the underlying logic plausible, that demand for software grows as the cost of producing it falls, while flagging the obvious conflict: "if your entire business depends on software engineering growing as a field, you have a strong incentive to believe that agents will have that effect." He grounds the doubt in a number, GitLab's stock falling from about 52 to about 26 dollars in a year, which makes a dramatic reorganization easier to read as a bet under pressure than as neutral foresight.

Why it matters

If you read vendor claims about agents reshaping engineering, this is a clean example of separating the argument from the arguer: the org change is real, but the company making it is the one with the most to gain from the thesis being true.