# Linear’s data: AI didn’t give your week back *Updated 2026-09-22 · Published 2026-08-17 · Matthew Blode* Linear's 2026 data report found time spent creating and triaging work rose across nearly every function, with nothing shrinking to make room. --- > **TL;DR:** Linear’s 2026 data report measured time spent creating, triaging, and commenting on work rising across nearly every function between June 2025 and June 2026. Chatting with AI and delegating to agents arrived on top, nothing shrank to make room, and planning time held flat. More work arrives, so the list has to stay calm. The story about AI at work is that it hands you time back. Linear went looking for that time in its own product data and didn’t find it. Its first data report, written by head of data Tim Qi, covers tens of thousands of teams building software in Linear, and its quietest finding is that coordination went up. ## What Linear measured Between June 2025 and June 2026, average minutes per user per month spent creating and triaging work rose in nearly every function. Engineering went from 24 to 28, which Linear puts at roughly 17%. Design went from 22 to 25, go-to-market from 27 to 31. Product was the only dip, 38 to 37. Commenting rose everywhere it was measured: engineering from 35 to 40, go-to-market from 49 to 55. Two caveats travel with every number here, and Linear states both. The data covers paid Linear workspaces only, so in Linear’s words it is “a picture of adoption inside our own customer base, not the market at large”. And AI used outside Linear is invisible to it, so a team running agents entirely in a terminal registers as a non-adopter. ## Nothing shrank to make room Chatting with AI and delegating issues to agents didn’t exist as categories of work a year earlier. By June 2026 chat with AI showed up at 2 minutes per user per month for engineering and 5 for product. Small numbers, but they were 0 the year before, and Linear’s own reading is blunt: “Nothing else shrank to make room, which suggests AI has landed on top of existing work rather than replacing any of it, at least so far.” Tim Qi’s closing note says it again: “teams are working more, not less.” Linear sells software that AI is meant to speed up, and it published the version of its data that says the week got fuller. That makes the finding easier to trust. ## More work, from more sources The share of users active on AI features more than doubled in every function between January and June 2026, with product climbing fastest, from 12% to 34%. In the week of 27 July 2026, agents and MCP created almost as many issues in Linear as people and integrations did: about 2.35 million against 2.36 million. Two years earlier it was fewer than one issue in a thousand. The one thing that held still was deciding what to build. Time spent on customer requests, docs, and projects stayed within 0 to +1 minute across every function measured. Linear reads that as AI changing how teams execute far more than how they decide what to build, with the caveat that planning practice varies widely between teams and much of it happens in conversation before it lands anywhere. So the volume climbed, the number of sources writing it climbed, and the human job of sorting it stayed put. The judgement about what deserves your attention didn’t get automated. It got surrounded by more input. ## What this means for your own list Everything above is measured inside an issue tracker, and the same pressure lands on the list you keep for yourself. If an assistant can file follow-ups the moment a call ends, and a script can file three more overnight, capture stops being the bottleneck. Deciding becomes the bottleneck. The hard part of a personal system used to be getting things into it, which is why so much productivity advice is about capture. When work arrives from a terminal, a browser, a phone, and an assistant, getting it in is solved. What’s left is the part no tool does for you. Done Bear sorts every open task into five Getting Things Done views: Inbox, Today, Upcoming, Anytime, and Someday. Triage is the work Linear measured going up, and five fixed destinations keep it quick, because there are only five answers. An assistant working over [MCP](/blog/what-is-an-mcp-task-manager), the [CLI, or the API](/automation) writes to that same list, so there is no second one to check. It is the same shape [as Things 3](/things-3-alternative), with sync and agent access. ## Where this is not an argument for Done Bear Done Bear doesn’t replace an issue tracker, and this report isn’t evidence that it should. Teams running sprints or cycles in Linear or Jira should stay there. Our [teams page](/teams) says so directly, and Linear’s data does not change it. The argument is narrower. Work arrives from more people and more agents than a year ago, planning time didn’t move to absorb it, and nothing was freed up. The list where you decide what to do next carries more traffic than it used to, so it needs to stay a list. The [free plan](/pricing) is enough to see whether five views hold up under yours. ## Sources - [Linear: How teams build](https://linear.app/data), written by Tim Qi and published August 2026. Every figure above comes from it: minutes per user per month by function between June 2025 and June 2026, the AI layer appearing at 0 to 5 minutes, planning holding within 0 to +1 minute, AI feature adoption by function, and issues created by agents and MCP against people and integrations in the week of 27 July 2026. --- Every Done Bear page is listed at https://donebear.com/llms.txt. Any page has a markdown copy at its URL plus `.md`. Start from the map rather than guessing a path.