Linear Declares Issue Tracking Dead: 25% of Issues Already Created by AI Agents — The End of Traditional Product Management
Linear's CEO announces 'Issue tracking is dead' on March 25, 2026. Linear Agent, Skills and Automations transform product management — 75%+ of enterprise workspaces already run agents.

"Issue tracking is dead." Three words published this morning by Karri Saarinen, CEO of Linear. And three numbers to prove it: 75% of enterprise Linear workspaces have agents installed. Agent work volume grew 5x in three months. And 25% of new issues are now created by agents. Not humans. Linear no longer presents itself as a tracking tool — but as "the shared product system that turns context into execution."
Issue tracking was institutionalized idle time
The classic product management model runs on a chain of handoffs. A PM defines the scope. They create an issue — a ticket describing a task to be done. The issue waits. A dev discovers it three days later. A prioritization meeting is scheduled. A negotiation begins.
Every step is a handoff — a transfer of responsibility between two people. And every handoff is dead time. Value sleeping inside a system.
Karri Saarinen puts it bluntly: "Issue tracking was built for handoffs. The process became the work." Complexity disguised itself as sophistication. Jira offers 500 customizable fields. Does that help ship faster? Or does it slow everyone down?
This is the Idlen philosophy applied to product management. Every handoff between a PM and a dev is an idle resource — activatable time that stays dormant. Linear just named this problem officially.
The 3 numbers that prove it's already happening
| Metric | Number | What it means |
|---|---|---|
| Enterprise workspaces with agents | 75%+ | Massive adoption already here |
| Agent work volume growth (3 months) | 5x | Exponential acceleration |
| Issues created by agents | 25% | 1 in 4 issues is no longer human |
These aren't projections. They're observations. In 2024, zero issues were created by agents. In three months of early 2026: one in four. The trajectory is clear. Within twelve months, the majority of triage issues — routine classification and assignment tasks — could be fully automated.
Karri Saarinen adds: "Agents are not mind readers. They become useful through context." The key isn't generic AI. It's AI connected to your product context.
Linear Agent + Skills + Automations: what changes today
Three products launch simultaneously.
Linear Agent is a native agent interface embedded in your product workspace. The agent analyzes user feedback, creates projects, issues and documents. It works within your existing product context — it knows your projects, your priorities, your history.
Skills are reusable workflows encoded as capabilities. They can be triggered manually via slash commands, or activated automatically by Linear when context warrants it. Skills capitalize on your team's learnings. It's the same concept as Figma Skills launched the day before — "Skills" is becoming the standard agent interface in 2026.
Automations (Triage) act the moment an issue enters the system. Linear refines, synthesizes or takes action automatically. Zero human intervention for routine triage — the classification and assignment of repetitive tasks. This automatic triage builds on the Triage Intelligence already deployed in Linear.
The metaphor: Linear becomes your first PM — one who never sleeps, never negotiates and never schedules meetings.
The roadmap: when Linear writes your code
Three features are announced for the coming months.
Code Intelligence: Linear understands your codebase, answers technical questions and helps with debugging. Code Diffs: code review directly inside Linear — an interface for humans and agents. Linear Coding Agent: Linear writes code and fixes bugs automatically, powered by frontier models (the most advanced AI models) and native Linear context.
When all three features go live, Linear will be the only tool that goes from user feedback to GitHub commit — without human intervention on routine tasks.
The natural question: where does Linear end and where does Cursor begin? Agentic product management and agentic development are converging toward the same point.
Jira vs Linear: the perfect irony
| Criteria | Linear | Jira (Atlassian) |
|---|---|---|
| Philosophy | Remove overhead | Absorb complexity |
| AI response | Native agent + Skills | Layoffs + pivot |
| Issue tracking | "Is dead" | Core business |
| Agent context | Full stack (feedback → code) | Partial |
| Coding Agent | Announced (roadmap) | Not announced |
| Auto triage | Live today | In development |
Earlier this month, Atlassian laid off 1,600 people — 10% of its workforce — to "pivot to AI." The same month, Linear announces agents are replacing humans on triage and issue creation.
Both are pivoting to AI. But Linear starts from a philosophy — "The best systems remove overhead so teams can focus on building." Atlassian starts from urgency — cutting costs. That's not the same thing.
In summary:
- Linear's CEO declares on March 25, 2026: "Issue tracking is dead" — built for handoffs, replaced by a system centered on context and agents.
- 75%+ of enterprise Linear workspaces have agents installed, agent volume grew 5x in 3 months, 25% of issues are created by agents.
- Launch of Linear Agent (native product context), Skills (reusable workflows) and Automations (automatic Triage).
- Announced roadmap: Code Intelligence, Code Diffs and Linear Coding Agent — Linear will write code automatically.
- Linear redefines itself: "the shared product system that turns context into execution."
Issue tracking was the institutional formalization of idle time. Every handoff between a PM and a dev, every issue waiting for triage, every prioritization meeting — that was value sleeping inside a system. Linear just named this problem and launched the tool that eliminates it. Yesterday, Figma opened its canvas to agents. This morning, Linear declared issue tracking dead. Monday, Lovable announced its acquisitions. In 72 hours, three of the most widely used product team tools changed their nature. It's no longer AI inside your tools. It's your tools becoming agents.


