Cursor launches a new agent-driven coding assistant for developers

As agentic coding becomes more widespread, the day-to-day work of a software engineer has grown strikingly more complicated. One engineer may now be responsible for supervising dozens of coding agents simultaneously, starting different processes and steering them as needed.

That creates a great deal to manage, and human attention has rapidly become the main bottleneck.

Cursor introduced a new tool on Thursday intended to bring more order to that complexity. Known as Automations, the new system lets users automatically start agents in their coding environment, with triggers that can come from a new codebase addition, a Slack message, or even a simple timer. In Cursor’s framing, the system is meant to help teams review and maintain all the new code generated by agentic tools — without requiring engineers to monitor dozens of agents at once personally

At its simplest, automation is intended to help engineers move beyond the “prompt-and-monitor” pattern that has defined much of agent-based software engineering so far. Rather than depending on a person to launch agents through a manual prompt, Cursor’s Automation framework makes it possible to start those agents automatically — while still bringing humans in whenever their involvement is actually needed.

“It’s not that humans are completely out of the picture,” Jonas Nelle, Cursor’s engineering chief for asynchronous agents, said in an interview. “It’s that they aren’t always initiating. They’re called in at the right points in this conveyor belt.”

One early illustration is Bugbot, an existing Cursor feature that the company now views as an early forerunner to the broader Automation system. Bugbot runs each time an engineer adds code and then reviews the new code for bugs and other issues. With Automations, Cursor has expanded the system to enable more comprehensive security audits and deeper review workflows.

“This idea of thinking harder, spending more tokens to find harder issues, has been really valuable,” said engineering lead Josh Ma.

Cursor estimates it now runs hundreds of automations per hour, far beyond basic code review tasks. The system is also used for incident response, where PagerDuty incidents can trigger an agent that immediately checks server logs through an MCP connection. Another automation provides weekly summaries of codebase changes in Cursor’s company Slack.

“In the abstract, anything that an automation kicks off, a human could have also kicked off,” Nelle said. “But by making it automatic, you change the types of tasks that models can usefully do in a codebase.”

The launch comes amid heavy competition in the agentic coding market, with both OpenAI and Anthropic rolling out notable updates to their agent-based coding tools over the past month.

Data from Ramp shows Cursor’s market share remaining stable since May, with about 25% of generative AI customers subscribing to Cursor in some form.

Even so, the broader expansion of the agentic coding space has continued to increase the company’s revenue sharply. Earlier this week, Bloomberg reported that Cursor’s annual revenue had climbed to more than $2 billion, doubling over the previous three months.