Two years ago, fewer than one issue in a thousand inside the project management tool Linear was created by AI. Today, AI writes just under half of everything created on the platform — and at the current pace, it will soon author more than people and integrations combined.
Those figures come from Linear's newly published "AI usage patterns in software teams" report, drawn from activity across its customer base of well over 100,000 paid users. It is among the clearest first-party pictures yet of how AI is actually changing day-to-day software work in 2026 — who uses it, how it reshapes where teams spend time, and whether it changes how much they ship. For more context on this story, see our ongoing latest AI developments.
AI Adoption More Than Doubled in Every Function
Between January and June 2026, the share of users active on Linear's AI features more than doubled in every job function measured — a breadth of adoption that surprised even the company's own analysts.
Product managers climbed fastest, from 12% to 34%. Even go-to-market staff — the function furthest from the codebase — went from 5% to 18%. The figures are based on 127,000 paid users who were active in both January and June 2026.
Company size, usually a reliable predictor of how fast organizations adopt new technology, "barely registers" in the data: adoption roughly tripled everywhere, from startups to enterprises.
Executives Are Using AI More Than Their Teams
The most striking jump in the entire report belongs to executives. CEOs at companies with 201 or more employees went from 9% active on AI features in January to 36% in June — the largest increase of any segment Linear measured.
The report's interpretation: the most senior leaders are learning the technology by using it rather than reading about it. For anyone following the workplace AI debate, that detail matters — the people ultimately responsible for AI strategy are increasingly hands-on users themselves.
Non-Engineers Are Shipping More Code
The boundary of who "ships code" is dissolving. The share of product managers attaching pull requests to their work rose from 3% to 10% in two years, and designers from 1% to 8%.
Engineering rose too, from 20% to 34%, and founders from 11% to 23%. Even go-to-market staff inched up from 1% to 3%. Linear cautions these are floors rather than ceilings, since it only counts pull requests in repositories connected to the platform — anyone shipping code outside that loop is invisible.
The pattern, as the report puts it: the people who used to describe a change increasingly ship it themselves.
Output Is Up 111% — But Planning Hasn't Changed
Pull requests opened per workspace are up 111% against a June 2024 baseline. Output held roughly level for the first year, then bent upward through 2026 as model quality and adoption climbed together.
Yet one thing stayed stubbornly constant: the time teams spend on customer requests, documents and project planning. That steadiness, Linear suggests, indicates AI has so far changed how teams execute far more than how they decide what to build.
The data shows AI chat and delegating issues to AI agents are new categories of work that "didn't exist a year ago" and now appear in every function's week, with product teams leaning in hardest. Notably, nothing else shrank to make room — suggesting AI has landed on top of existing work rather than replacing any of it, at least so far.
The scale behind those numbers is substantial. Linear's user base grew from 54,300 paid users in June 2025 to 89,000 a year later, and the time users spend chatting with AI and working on agent-delegated issues now shows up across all of them — averaging 2 minutes per user for engineers and 5 minutes for product managers on AI chat alone. Small individually, but as the report notes, these categories registered zero minutes two years ago.
Read the Fine Print
Linear is transparent about its limitations, and they matter. The company can only see AI usage that happens inside its own platform — the picture reflects adoption among its customers, not the market at large. Model companies and coding toolmakers publish token usage and code volume, but as the report notes, that captures only one layer of the work.
There are also measurement caveats: the report counts pull requests opened rather than merged, and an opened pull request says nothing about the value of the change. Job titles are normalized with some error at the edges, and the executive analysis relies on third-party company-size enrichment, covering fewer workspaces than the rest of the report.
A Fixed Point for 2026
Linear describes the report as "a fixed point for where AI-assisted product development stands in 2026, and something to measure the next edition against."
That fixed point is remarkable nonetheless. In eighteen months, AI inside one of the industry's most widely used development tools went from writing one issue in a thousand to nearly half of all of them — while the humans moved up the stack, describing less and shipping more. The next measurement, a year from now, will show whether the planning layer finally moves too.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
