Cursor, the AI-powered coding tool that shook up the developer world, has announced its next major step: a web app that lets you task and manage an army of AI agents to handle complex coding tasks directly from your browser.
Introduction: A new step toward autonomous programming
Anysphere, the company behind the popular AI-powered code editor Cursor, launched a new web app on Monday that allows users to manage a network of “coding agents” directly from their browsers. This launch represents a major strategic step for the company as it expands beyond its core product, the integrated development environment (IDE), and delves deeper into the world of autonomous agent-driven experiences.
What are “coding agents”?
To understand the significance of this announcement, one must first understand what “coding agents” are. Unlike AI tools that suggest snippets of code as you type, coding agents are artificial intelligence systems that can solve entire programming tasks independently without continuous user supervision. You can ask a coding agent, for example, “Add Google sign-in to my app,” and the agent will proceed to write, test, and debug the necessary code on its own.
The evolution of Cursor’s strategy
In May, Cursor launched its background-running agents. In June, it launched an integration with Slack, allowing users to assign tasks to these agents simply by sending a message. Now, with the new web app, Cursor users can send natural language requests via the browser—on desktop or mobile devices—to assign tasks to agents, such as writing new features or fixing bugs in their codebase. The web app also allows users to monitor the agents’ work, view their progress, and merge completed changes into the core codebase.
“Removing barriers” for developers
Andrew Milich, head of product engineering at Cursor, says these steps are aimed at “removing barriers” for users. Instead of a developer being restricted to opening a complex development environment to start a task, they can now initiate the process through a simple Slack message or via the web app, allowing the smart agent to take the first crack at it. If the agent cannot complete the task, the developer can seamlessly transition to the IDE to finish where the agent left off.
Commercial success supporting expansion
Cursor’s strategy appears to be paying off. The company announced last month that its annual recurring revenue has exceeded $500 million, largely driven by monthly subscriptions. It also stated that more than half of Fortune 500 companies now use its tools, including companies like Nvidia, Uber, and Adobe. To capitalize on this growth, the company recently launched a $200-per-month professional subscription tier.
The race for reliable coding agents
Cursor is not the first to launch coding agents, but the company says it was careful not to release “demoware”—products that look good in presentations but fail in practice. The Cursor team now believes that AI reasoning models are advancing enough to make coding agents practically viable. The company’s CEO expects coding agents to handle at least 20% of a software engineer’s work by 2026.
Conclusion: The future of collaborative programming
The launch of Cursor’s web app marks a milestone in the evolution of developer tools. It reinforces the idea that the future of coding will not just be human writing code, but rather a dialogue and collaboration between human developers and autonomous smart agents. With the ability to manage this collaboration from anywhere at any time, Cursor is one step closer to realizing the vision of seamless and efficient programming that developers have long dreamed of.
