With the proliferation of autonomous coding methods, the professional existence of a software engineer has grown remarkably intricate. It’s now common for a sole engineer to supervise numerous coding agents simultaneously, initiating and directing diverse operations as needed.
This presents a substantial volume of data to monitor, leading the focus of human engineers to swiftly emerge as the primary constraining factor.
On Thursday, Cursor unveiled an innovative instrument designed to manage that disarray. This new system, named Automations, furnishes users with a method to autonomously initiate agents directly within their coding milieu, activated by events such as a fresh inclusion in the codebase, a Slack message, or a basic temporal trigger. As articulated by Cursor, it serves as a mechanism to examine and upkeep all new code generated by autonomous utilities, thereby circumventing the need to monitor scores of agents concurrently.
At its fundamental essence, Automations enable engineers to transcend the “prompt-and-monitor” paradigm that characterizes most agent-based development. Rather than initiating agents via manual commands, Cursor’s Automation framework facilitates autonomous agent deployment, engaging human operators only at opportune moments.
“It’s not that humans are entirely excluded,” Jonas Nelle, Cursor’s engineering chief for asynchronous agents, conveyed to TechCrunch during an interview. He further elaborated, “They don’t consistently instigate actions; instead, they are engaged at crucial junctures within this operational assembly line.”
An initial illustration is Bugbot, an established Cursor functionality that the team views as a forerunner to the more extensive Automation framework. The Bugbot system is activated each instance an engineer introduces a modification to the codebase, scrutinizing the fresh code for defects and other discrepancies. By leveraging Automations, Cursor has succeeded in augmenting that functionality to encompass more comprehensive security examinations and in-depth assessments.
“The concept of more intensive deliberation, dedicating additional computational resources to uncover more complex problems, has proven exceptionally beneficial,” stated engineering lead Josh Ma.
Techcrunch event
San Francisco, CA
|
October 13-15, 2026
Cursor projects it performs numerous automated processes hourly, extending well beyond mere code examination. This system is additionally employed for crisis management, where PagerDuty events can trigger a bot capable of instantly retrieving server records via an MCP interface. Furthermore, a distinct automated process provides weekly synopses of modifications to the codebase within Cursor’s corporate Slack channel.
“Theoretically speaking, any action an automation initiates, a human could equally have commenced,” Nelle remarked. “However, by automating it, you fundamentally alter the nature of assignments that models can productively perform within a code repository.”
This novel system is introduced amidst fierce rivalry within the autonomous coding domain, as both OpenAI and Anthropic have implemented substantial enhancements to their respective autonomous development utilities over the preceding month.
Data from Ramp indicates Cursor’s market penetration has remained stable since May, with approximately a quarter of clients utilizing generative AI engaging with Cursor’s offerings in various forms.
Nevertheless, the collective advancement of the autonomous coding sector has sustained the firm’s income augmentation at an astonishing rate. Earlier this week, Bloomberg disclosed that Cursor’s yearly income had ascended beyond $2 billion, experiencing a twofold increase during the preceding quarter.
{content}
Source: {feed_title}