Coinbase's head of engineering for Base App, Chintan Turakhia, issued an unusual directive earlier this year: delete your IDEs and write zero lines of code manually. The team complied.

The context is Linear's newly launched Linear Agent, announced March 24. Linear โ€” the product management and issue-tracking tool used heavily by engineering teams โ€” has repositioned itself as a platform where AI agents do the procedural work of software development. The system connects customer feedback, tickets, strategic context, and codebase into a unified workspace that agents can both read and act on.

According to Linear's CEO Karri Saarinen, coding agents are now installed in more than 75% of Linear's enterprise workspaces. Over the past three months, agent-completed work grew 5x, and agents authored nearly 25% of new issues. For Coinbase's Base App team, development has become continuous โ€” not because engineers are working around the clock, but because agents are.

Linear Agent is accessible directly inside the app, in Slack, and in Microsoft Teams. It can synthesize backlogs, surface relevant feature requests, and draft project specs from customer requests in minutes. Alongside it, Linear shipped Skills (reusable agent workflows triggered by slash command) and Automations (agent actions triggered when issues enter triage).

Code Intelligence โ€” which will allow agents to answer questions about and debug the codebase โ€” is coming soon to Business and Enterprise plans.

The shift is significant: traditional issue tracking was built around handoffs between roles. Linear is betting the next model is context-to-execution, with agents collapsing the distance between intent and shipping code.