What broke when we tried to just use Cursor
Every engineering org I've watched install Cursor gets the same 20% bump. None of them stay there. Here's what actually moved the needle — and why it wasn't the editor.
Replace this body with the real essay before public launch. This stub ships so the
/essaysindex isn’t empty during development, so the homepage “Writing” section has three cards to render, and so the OG image generator has a slug to emit.
Some placeholder body text follows so reading-time computes above 1 minute. Delete everything below before launch.
The first week after Cursor landed team-wide, velocity jumped. Not by a lot — 15%, maybe 20%. Tickets closed faster, pull requests landed faster. Standup felt lighter.
By week six the graphs had flattened. Same team, same tool, same 20% lift over the pre-Cursor baseline. The ceiling wasn’t the tool. The ceiling was everything around it.
What we eventually changed took six months to map and about the same again to roll out. The short version is this: the pattern isn’t about making the developer faster at typing. It’s about making the delivery system legible to an agent that doesn’t think in your team’s implicit context.