This is a dashboard of adoption and output metrics from Linear's customer base (tens of thousands of software teams) spanning June 2024 to August 2026, tracking how AI features propagate across roles and company sizes, and whether they correlate with measurable output shifts. The core claims: AI adoption roughly tripled across every function and company size in six months; AI now authors ~45% of all issues created in Linear; non-engineers (PMs, designers) are shipping significantly more pull requests; and teams with coding agents nearly tripled their weekly PR volume over two years (21→65 PRs/week) while traditional teams stayed flat (8→10). The report explicitly flags its own blindspots—it can't see AI usage outside Linear, only measures motion (PRs opened, not merged) not value, and observes that existing work-time didn't shrink when AI arrived, suggesting additive rather than substitution work. The timing window and aggregation method appear sound (monthly cohorts, trailing 30-day windows for adoption), though the source doesn't explain whether the "coding-agent teams" cohort self-selected for higher capability before agents existed. One sharp question: the report shows agent-created issues climbing from near-zero to 2.4k/week by August 2026, yet doesn't break down whether those issues are being triaged/closed at different rates than human-created ones, which would hint at quality differences.