Linear's Agents Now Create Half Its Work. A Year Ago It Was 3%.

Linear disclosed that agents are now installed in 95% of its paid workspaces, with the share of work items agents create jumping from 3% to 50% over twelve months, and issues with attached pull requests up sevenfold since January 2026 — meaning agents are increasingly completing work, not just logging it. The company raised $99 million in secondary funding at a $2.5 billion valuation, double its prior Series C mark, while staying cashflow positive at 177% net revenue retention on $100M+ ARR. Jason Lemkin’s read, contrasting Linear against competitors, sharpens the pattern: Atlassian reports 4x growth in agent-generated work items with MCP server monthly active users nearing 1 million (adopters expanding seats twice as fast as non-adopters, while 98% still use the Jira UI the same month — agents augmenting, not replacing, human usage); monday.com’s growth has slowed to 16-17% guidance with a 20% workforce cut; Asana is guiding 8-9% growth after acquiring StackAI.

Lemkin closes with three questions he says every B2B product leader should be tracking — what share of records agents now create, whether agents carry distinct identities inside the product, and whether pricing scales with work volume instead of headcount — and that last question isn’t hypothetical for RevOps tooling. Lemkin made the broader pricing case himself a few days earlier: per-seat pricing is dying for three converging reasons, only one of which is AI, and the Vertice SaaS Inflation Index has run 12-16.4% through 2026 against roughly 2.7% general inflation, while Zylo’s SaaS Management Index shows average enterprise SaaS spend hitting $55.7M a year even as app-portfolio counts stayed flat. Read against Linear’s own numbers, the pricing question isn’t abstract: if half a workspace’s output is agent-generated, a seat count stops measuring what a customer is actually paying for.

For sales and RevOps leaders evaluating their own tool stack, Linear’s disclosure is a rare instance of a vendor putting exact percentages on agent adoption inside a live product — a benchmark worth checking your own tools against.