AI Now Authors Half of All Issues: Linear's Data Reveals How Software Teams Really Use AI in 2026

AI coding tools get a lot of hype, but most of what we know about their impact comes from token counts and code volume metrics. Linear just changed that. With tens of thousands of teams building software inside its platform every day, Linear has a unique vantage point: they can see the entire workflow from the first issue to the pull request that closes it. Their first-ever report on AI usage patterns in software teams — covering 127,000 paid users — reveals something more nuanced than the usual "AI is transforming everything" narrative.

The headline finding: AI now authors nearly half of all issues created in Linear. Two years ago, fewer than one issue in a thousand was AI-generated. Today, agents and MCP clients create almost as many issues as humans and integrations combined. At the current trajectory, AI will soon author more issues than every other source put together.

AI Adoption Spread to Every Function — Not Just Engineering

Between January and June 2026, the share of users actively using AI features more than doubled in every function. Engineering went from 12% to 30%. Product climbed fastest of all, from 12% to 34%. Even go-to-market teams — the function furthest from the codebase — went from 5% to 18%. The pattern is too broad to be a labeling artifact.

Here is the adoption breakdown by function:

• Founders: 14% → 30% (+16 points)
• Engineering: 12% → 30% (+18 points)
• Product: 12% → 34% (+22 points — the fastest climb)
• Design: 6% → 22% (+16 points)
• Go-to-market: 5% → 18% (+13 points)

What makes this remarkable is that company size barely matters. AI adoption roughly tripled everywhere, from one-person startups to thousand-person enterprises. The technology that was supposed to favor early-stage agility has spread evenly across the entire spectrum of company size.

CEOs Are Using AI More Than Anyone

Perhaps the most surprising finding: executives are personally active on AI at rates that match or beat their teams. CEOs at companies with 201+ employees went from 9% to 36% in six months — the single largest jump in the entire report. CTOs at large companies went from 11% to 35%. CPOs at small companies went from 11% to 36%.

This suggests something important about how AI is diffusing through organizations. The most senior leaders are not just greenlighting AI initiatives — they are learning the technology by using it themselves. They are not reading about AI in strategy decks. They are opening the tool, typing prompts, and forming opinions based on first-hand experience.

The Issue Creation Revolution

The most striking chart in Linear's report tracks who creates issues over time. In June 2024, AI-created issues were essentially zero — fewer than one in a thousand. By August 2026, agents and MCP clients are creating over 2,400 issues per week, compared to roughly 2,480 from people and integrations. AI is on track to become the majority author of issues within weeks.

The growth curve is not linear — it is exponential. The inflection point came in mid-2025, when AI issue creation jumped from single digits per week to hundreds. By early 2026, it crossed a thousand. By mid-2026, it crossed two thousand. The trajectory shows no sign of plateauing.

Coding Agents Triple Output — But Only for Teams That Use Them

Linear tracked a fixed cohort of teams from June 2024 to June 2026. Teams that connected a coding agent roughly tripled their weekly pull requests, from 21 to 65 per week. Teams without a coding agent went from 8 to 10 — essentially flat.

The catch: these coding-agent teams were already higher-output before the agents existed. They were the teams predisposed to adopting new tools and shipping aggressively. The levels are not directly comparable, but each cohort against its own baseline tells a clean story. Nearly all the growth in output sits on the agent side.

Overall, pull requests per workspace are up 111% over two years. Output held roughly level for the first year, then bent upward through 2026 as model quality and adoption climbed together.

Non-Engineers Are Shipping Code

The share of product managers attaching pull requests in Linear rose from 3% to 10% in two years. Designers went from 1% to 8%. Even go-to-market teams went from 1% to 3%. Linear notes these are floors, not ceilings — they only count PRs in repositories connected to Linear, so anyone shipping outside that loop is invisible.

The implication is significant: the people who used to describe a change increasingly ship it themselves. The boundary between "I want this feature" and "I built this feature" is blurring, and AI is the tool making it possible.

The Jevons Paradox of AI Productivity

Here is the finding that challenges every ROI calculation about AI tools: teams are not saving time. Time spent on existing tasks in Linear held steady while AI usage appeared as an entirely new layer of work. The overall time spent on product development is going up, not down.

Engineering time on creating and triaging issues rose from 24 to 28 minutes per user per month. Commenting time rose from 35 to 40 minutes. Founders showed the largest swings — up 17 minutes on creation and 26 minutes on commenting. A new category of work — chatting with AI and delegating issues to agents — appeared out of nowhere and now shows up in every function's week.

Linear explicitly calls this a Jevons paradox: when a resource becomes more efficient to use, consumption of it goes up rather than down. AI makes it cheaper to create issues, write code, and communicate — so teams do more of all of it. The productivity gains show up as more output, not less effort.

What Planning Time Tells Us

One metric that did not move: planning time. Time spent on customer requests, docs, and projects held steady across all functions. Linear's interpretation is measured: AI has so far changed how teams execute far more than how they decide what to build. The strategic layer — deciding what to prioritize, what to build, what to skip — remains a human activity, at least for now.

What This Means for Software Teams

Linear's data offers several actionable takeaways for teams navigating the AI transition:

• AI adoption is not optional — it is happening across every function and company size at similar rates. Teams that resist are not gaining a quality advantage; they are falling behind on output velocity.
• Coding agents are the highest-leverage AI investment — teams with coding agents tripled their PR output while teams without barely moved. If you have not connected a coding agent to your workflow, this is the single biggest lever available.
• Expect more work, not less — AI adds a new layer of work rather than replacing existing work. Plan for increased issue volume, more comments, and more coordination overhead. The gain is in output, not in time saved.
• Role boundaries are blurring — PMs and designers are shipping code. This is not a threat to engineers; it is a expansion of who can contribute. Teams that embrace this will move faster than teams that gatekeep.
• Planning remains human — AI has not yet changed how teams decide what to build. The strategic layer is still the hardest to automate, and it may be the most valuable human skill to develop.

The Bottom Line

Linear's report is the most concrete data we have on how AI is actually reshaping software development — not in theory, but in practice, across tens of thousands of real teams. The story is not the simple productivity gain that vendors sell. It is a fundamental restructuring of how work flows through teams: AI authors the issues, agents write the code, non-engineers ship changes, and everyone spends more time coordinating it all.

The teams winning with AI are not working less. They are working differently — and shipping significantly more. Whether that increased output translates to meaningful business results is the question Linear plans to investigate next. But the correlation between AI adoption and acceleration is undeniable. The teams that are not adopting are not standing still. They are falling behind.