Agents File the Tickets, Humans Approve Them at 10 AM

What people were saying

Quiet day, no Linear team posts, and the dominant theme was one Linear presumably likes: Linear as the place agents write things down.

The most-engaged post of the day wasn't really about Linear at all. @altryne spent 632 likes' worth of attention comparing two coding models, and the detail that mattered here was almost incidental: the model he likes is "architecting, fixing on the spot, updating Linear tickets" while the other one grinds through a settings page burning tokens. Ticket updates have quietly become a benchmark category, which is a strange and slightly flattering place for an issue tracker to end up.

The purest version of this came from @robinebers, who wired Cursor's new projects up to his Linear issues and had it triage everything through an Eisenhower matrix overnight. Ten proposed tasks at 10 AM for him to approve, and on the first run: 19 issues closed (confirmed in code), 27 updated, 4 created to merge duplicates, 5 PRs. He called it wild, which seems fair. @majroth built something adjacent, a morning brief where Codex reads email, calendar, Figma comments, Slack and Linear issues and proposes todos it can then assign back to itself. And @TSRviral relayed a Cursor demo where a cloud agent took a single ticket against Grafana's 2M-line codebase and shipped a feature with a walkthrough video, with Linear listed alongside Slack as a place to fire agents from.

The praise

Two separate design compliments, unprompted. @jaredjames_, launching a social scheduling tool, went on a small tear about everything looking like the same component library now and named Linear as the last product that made the mundane beautiful on purpose. @AntonHong_ simply shipped something and credited Linear for the inspiration.

On the workflow side, @robinebers wrote a widely-shared post about shipping fast and understanding maybe a quarter of your own architecture, in which the load-bearing habit was throwing every known problem into a Linear ticket and moving on. He has since deleted 100k+ lines of slop, so the tickets apparently worked.

The pushback

Nothing aimed at Linear directly, but two replies poked at the agent premise. One reader under @altryne noted that silent Linear updates from an agent really want an audit log before anyone runs this in production. Under @robinebers, someone made the sharper version: the debt you ticketed is fine, it's the debt you never noticed that compounds. Also, in a UI-scoring demo from @dnzbuyuktas, an AI judged Linear a 66, behind Notion, Stripe and Vercel. Take that as you like.