Linear, the project management platform used by tens of thousands of software teams, published usage data this week showing something that cuts against the usual pattern of enterprise software adoption: the people at the top are moving faster than the people doing the work. Among CEOs at companies with more than 200 employees, the share active on Linear’s AI features in a 30-day window rose from 9 percent in January 2026 to 36 percent in June, a 27-percentage-point jump that outpaced every other role or company-size cut in the report. Linear’s head of data, Tim Qi, framed it as senior leaders “learning the technology by using it rather than reading about it.”
That framing matters because it inverts the standard story about enterprise AI rollout, where line workers experiment first and executives approve budgets later. Founders and CTOs at large companies also more than doubled their adoption over the same period, and Linear found no consistent gap by company size at all. Workspaces with more than 1,000 employees and workspaces under 50 employees each saw adoption roughly triple, converging in the same 23 to 27 percent range by June.
The adoption curve shows up downstream in what actually gets built. AI authored roughly one issue out of every thousand logged in Linear back in 2024. By this summer, AI accounted for close to half of all new issues created in the product, and Linear says that share is on track to overtake the combined total from people and integrations. Workspace pull-request volume has climbed 111 percent since the June 2024 baseline Linear uses for comparison, with the increase accelerating sharply through 2026 as coding-agent adoption spread.
The clearest split in the data sits between teams that connected a coding agent and teams that did not. Weekly output for agent-connected teams climbed from 21 pull requests to 65 across two years, a threefold increase. Teams running without a coding agent inched from 8 to 10 over the same stretch. Linear notes these cohorts were not equivalent to begin with, since agent-adopting teams already shipped more before agents existed, so the comparison measures acceleration within each group rather than agents against non-agents head to head.
One number complicates the productivity story: planning time did not move. Only one metric in the report refused to budge: the hours teams logged per month on customer requests, documentation, and project planning were essentially unchanged from June 2025 to June 2026, even as AI usage, issue volume, and pull request output all climbed together. Time spent chatting with AI or delegating work to agents shows up as a new category in every function’s week rather than replacing existing tasks. The hours automation frees are not becoming hours saved. They are becoming more work.
Two caveats belong alongside Linear’s own. This is Linear’s own customer base, not the software industry at large. Teams that pay for a modern project management tool and connect coding agents to it already skew toward the technical, well-resourced end of the market, so the numbers describe the leading edge of adoption rather than the average team. Linear also counts pull requests opened, not merged or reviewed, so the 111 percent output jump measures activity, not verified value shipped.
For engineering leaders benchmarking their own AI rollout, the useful comparison is not whether the team uses AI at all but whether a coding agent is actually wired into the pipeline. That single variable is what separated a tripling in weekly output from a low single-digit gain in Linear’s data, and it is the one worth checking before the next planning cycle.
Linear published this usage analysis, authored by head of data Tim Qi, in its “How Teams Build” report (Edition 01, 2026).
