Cursor introduces a fresh AI agent experience in which he can face Claude Code and Codex

The cursor was announced on Thursday the launch of Cursor 3, a fresh product interface that allows users to launch AI coding agents to perform tasks on their behalf. The product, which was developed under the code name Glass, is Cursor’s answer to agent-based coding tools such as Anthropic’s Claude Code and OpenAI’s Codex, which have gained recognition among millions of developers in recent months.

“Over the last few months, our profession has completely changed,” Jonas Nelle, one of Cursor’s engineering heads, told WIRED. “Many of the products that attracted Cursor are no longer as important in the future.”

Cursor increasingly competes with leading AI labs for developers and enterprise clients. The company pioneered one of the first and most popular ways for developers to code using AI models from OpenAI, Anthropic and Google, making Cursor one of these companies’ largest AI clients. However, over the past 18 months, OpenAI and Anthropic have released their own agent coding products and started offering them as highly subsidized subscriptions, putting pressure on Cursor’s business.

While Cursor’s core product allows developers to code in an integrated development environment (IDE) and employ an AI model for assistance, fresh products like Claude Code and Codex focus on enabling developers to offload entire tasks to an AI agent – sometimes running multiple agents at once. Cursor 3 is a startup version of the “agent-first” coding product. According to Nelle, the product is optimized for a world where developers spend their days “talking to different agents, checking them out and seeing what work they did,” rather than writing code themselves.

Cursor launches the fresh agent coding interface in an existing desktop application, where it will run alongside an IDE. In the center of the fresh Cursor window is a text box where users can type in natural language the task they want the AI ​​agent to perform – it looks more like a chatbot than a coding environment. Press Enter and the AI ​​agent will start running without requiring the developer to write a single line of code. In the sidebar on the left, developers can view and manage all AI agents running in Cursor.

What sets Cursor 3 apart from the desktop apps for Claude Code and Codex is that it integrates an agent-based product with Cursor’s AI-powered development environment. In the demo, Cursor 3’s other co-head of engineering, Alexi Robbins, showed WIRED how users can ask a cloud agent to run a feature and then review the code generated locally on their computer.

Nelle and Robbins argue that it doesn’t matter what interface developers spend their time on – they just want people to employ Cursor.

Competition with AI labs

Last week I visited the Cursor office in the North Beach neighborhood of San Francisco. According to reports, the startup is raising fresh capital of approx $50 billion valuation— almost twice as much as the financing estimate round last fall and transformed into an ancient cinema. Cursor employees used to throw their shoes into a pile by the door upon entering, but now there’s a row of immense shoe racks, signaling one of the company’s ways of growing.

However, Cursor still feels like a startup. Employees tell me that’s part of the appeal of working there; the company can ship quickly and doesn’t feel too corporate. But as it tries to catch up with Anthropic and OpenAI in the agent coding race, this reluctance may not be enough. This battle – the one to create the best AI coding agent – may be the most capital-intensive chapter in Cursor’s history to date.