A year ago, agents created just 3% of the issues in Linear. Now, it’s more than half. Cristina Cordova, COO of Linear, explains how the company’s go-to-market and product strategy changed as agent-driven workflows—and the budgets behind them—became central to the business.
Linear’s MCP server, shipped in May 2025, was its entry into the agent orchestration space. Since then, the company has integrated with tools like Codex, Cursor, and GitHub Copilot, and built its own Linear Agent. The result: over 50% of tickets are now generated by agents, not humans. ARR has crossed $100M, with over 40,000 paying customers and 177% net revenue retention. The company’s approach to AI is driving both usage and revenue, without the traditional enterprise sales playbook.
Agent Workflows Drive Both Usage and Retention
Cristina watches two metrics: agent usage and app usage. If agents create more work but people don’t return to the app, the product risks becoming a commodity. At Linear, both lines are up—users still return to coordinate and review work, even as agents automate much of the manual filing, labelling, and triaging. The company sees its product as a development system built around agent orchestration, not just an issue tracker with AI bolted on.
The Impact of AI Budgets on SaaS Growth
One major change: companies now have explicit AI budgets. This means products that fit into the new line item can bypass traditional procurement fights. Some of Linear’s competitors have hit $100M ARR quickly, but often with a few massive contracts; Linear’s growth is spread across tens of thousands of customers, mostly on seat-based pricing. Cristina notes buyers are signing shorter, one-year deals and paid pilots, reflecting uncertainty about what AI tools they’ll need long-term.
Rethinking Enterprise Sales and Support
Three weeks after joining, Cristina scrapped Linear’s “move upmarket to enterprise” plan, focusing instead on growth-stage companies—mirroring Stripe’s early strategy. Linear has never hired a single SDR, believing the qualification call is a waste of time. Instead, AEs do their own outbound, supported by tooling to keep outreach personal. The company also doesn’t do forward deployed engineering; the product should be easy enough to adopt without deep integrations. Support remains human-in-the-loop: AI drafts replies, but a person sends them.




