The rise of agentic coding has made the working life of a software engineer dazzlingly complex. A single engineer might now oversee dozens of coding agents at once, launching and guiding different processes as necessary. It is a lot to keep track of, and human attention has quickly become the limiting resource.
To help manage that chaos, Cursor is launching a new tool called Automations. This system allows users to automatically launch agents within their coding environment. These automations can be triggered by a new addition to the codebase, a Slack message, or a simple timer. Cursor describes it as a way to review and maintain all the new code created by agentic tools without having to track dozens of agents manually.
At its core, Automations is a way for engineers to move beyond the “prompt-and-monitor” dynamic of most agent-based engineering. Instead of launching agents with a human prompt, the framework lets agents launch automatically and loop humans in only when they are needed.
Cursor’s engineering lead for asynchronous agents, Jonas Nelle, explained the shift. He said it is not that humans are completely out of the picture, but that they are not always initiating. They are called in at the right points in the process.
An early example is Bugbot, a longstanding Cursor feature seen as a predecessor to the broader Automation system. Bugbot is triggered every time an engineer adds to the codebase, reviewing the new code for bugs. Using automations, Cursor has expanded that system to handle more involved security audits and reviews.
Engineering lead Josh Ma noted that this idea of thinking harder and spending more computational resources to find difficult issues has been very valuable.
Cursor estimates it currently runs hundreds of automations per hour, extending far beyond simple code review. The system is also used for incident response, where a PagerDuty alert can initiate an agent to immediately query server logs. A separate automation provides weekly summaries of codebase changes on the company’s Slack.
The new system arrives during intense competition in the agentic coding space. Both OpenAI and Anthropic have made significant updates to their agentic coding tools in the past month. According to one market data report, Cursor’s market share has held steady since May, with roughly 25 percent of generative AI clients subscribing to Cursor in some capacity.
Despite the competition, the overall growth of agentic coding has driven the company’s revenue upward at a stunning pace. A recent report stated that Cursor’s annual recurring revenue has grown to more than two billion dollars, doubling over the past three months.