‘Issue tracking is dead’: Linear CEO explains why the company is betting on agents

Software teams have spent decades organising work through tickets, backlogs and status updates. Linear, one of the stalwarts of the modern project management space, now says that model is reaching its limits.

The company last week introduced Linear Agent, a new system designed to act on context across a workspace rather than wait for tasks to be manually created and assigned. At the same time, co-founder and CEO Karri Saarinen has set out a broader position: issue tracking, as a central organising layer, “is dead.”

Linear Agent replaces handoffs with context

Linear Agent is built into the company's existing product and works across conversations, issues and project data. Instead of relying on users to define work step by step, the agent interprets context and takes action – creating issues, updating them, and handling routine coordination. It is accessible via a chat window in the bottom-right of Linear's desktop and mobile apps, by mentioning @Linear in any comment or reply, or via Slack and Microsoft Teams.

When starting a new project, for instance, a team no longer needs to manually trawl through past feature requests to find what is relevant. A user can simply ask Linear Agent to find related issues, group them by relevance, and pull the right ones in — then extract common requirements across customer requests and scope out a starting point for a spec.

When a team is staring down a cluster of related issues, a user can prompt the agent to make a plan, and it will generate suggested next steps based on the surrounding context.

That behaviour reflects a shift described by Saarinen in an open letter published alongside the launch.

Issue trackers, he argues, were designed for a model where work moved between people through explicit handoffs. Systems existed to capture those transitions: someone writes a ticket, someone else picks it up, and progress is tracked along the way. Over time, that process became the work.

"The best systems remove overhead so teams can focus on building,” Saarinen wrote.

That structure made sense when coordination was the bottleneck. It becomes less central when systems can interpret context directly.

And so Linear Agent follows that approach. It treats context—conversations, feedback, and code changes—as the source of truth, generating or updating issues as needed rather than relying on them as the starting point.

At the same time, Linear is also introducing agent "skills" — reusable workflows that can be saved and triggered on demand. When a conversation with the agent produces a useful result, that sequence can be codified and run again later, either manually via a slash command or automatically when Linear judges it to be relevant. A team that regularly splits large issues into sub-tasks, for instance, can save that as a skill and apply it with a single prompt rather than repeating the process from scratch each time.

What replaces issue tracking

Saarinen doesn’t claim that “issues” themselves are gone, his view is more that their role is changing. Issue tracking systems impose structure on work that increasingly does not need it. They require users to translate context into tickets before anything can happen – adding friction as the volume of information flowing across tools and channels continues to grow.

His argument is that this translation step is becoming unnecessary.

“Agents are not mind readers – they become useful through context,” he said. “Customer feedback, internal ideas, strategic direction, decisions, and code all need to be captured in a system that humans and agents can work from together.”

That system, he argues, should be able to interpret intent, route work to the right actor, and escalate when needed, keeping execution moving rather than trapping teams in process. And Linear is being rebuilt around that model.

Linear’s agentic roadmap

Looking a little further ahead, Linear’s roadmap extends that approach further into the development process. This will include Code Intelligence, which brings the agent’s understanding into the codebase, allowing teams to query systems, diagnose functionality, and design technical specifications.

Elsewhere, Code Diffs – also “coming soon” – is a review interface designed for both humans and agents, while a new Linear Coding Agent will write code and fix bugs using the same contextual model.

Linear Agent is now available in public beta across all plans. Core functionality, including agent chat and skills, is included as part of the base product, while higher-compute features such as automations and upcoming code capabilities are limited to Business and Enterprise tiers.

During the beta period, all features are available at no additional cost, with usage-based pricing expected for more intensive capabilities as the product matures.