Cursor announces major update to AI agents as coding tool battle heats up

Cursor announces major update to AI agents as coding tool battle heats up

Cursor, the innovative AI startup, revealed significant enhancements to its coding agents on Tuesday, aiming to solidify its position in a rapidly evolving market dominated by competitors like Anthropic, OpenAI, and Microsoft. As the company strives to maintain its edge and attract new users, it reported to CNBC that these updated agents represent a major leap forward in their capabilities. With a valuation soaring to $29.3 billion, Cursor also disclosed that it had surpassed $1 billion in annualized revenue by November. Having entered the AI coding sector early, Cursor now faces an influx of competing solutions. The updated AI agents are designed to autonomously complete various tasks for users and have gained immense popularity over the past year as advancements in AI models have emerged. Software developers, in particular, have embraced these agents for generating, editing, and reviewing code. The new features allow agents to test their modifications and document their processes through videos, logs, and screenshots. Users can activate these agents from multiple platforms, including the web, Cursor's desktop application, mobile devices, Slack, or Microsoft's GitHub. Moreover, the agents are able to operate within full development environments on cloud-based virtual machines, ensuring they do not compete for local resources on a developer's laptop. Alexi Robbins, co-head of engineering for Cursor's asynchronous agents, explained that this setup enables users to run numerous tasks simultaneously, drastically increasing productivity: "Instead of having one to three things that you're doing at once that are running at the same time, you can have 10 or 20 of these things running. You can have really high throughput with this." In the competitive landscape, Anthropic's Claude Code is reportedly generating over $2.5 billion in run-rate revenue, while OpenAI's Codex has attracted more than 1.5 million weekly active users. Microsoft CEO Satya Nadella mentioned that GitHub Copilot boasts over 26 million users. Founded in 2022, Cursor gained significant attention following the launch of its initial AI coding product. The latest updates enable developers to delegate more complex tasks to the agents, allowing them to focus on higher-level decisions rather than tedious coding and editing chores. Jonas Nelle, co-head of engineering for asynchronous agents, remarked that this evolution signifies a shift in how developers interact with AI, stating, "They’re not just writing software, writing code, they're sort of becoming full software developers." Internal testing of the agents has already shown promising results, with approximately 35% of Cursor's pull requests now generated by these agents operating on their virtual machines. Nelle emphasized the transformative potential of these tools: "You as an individual can do so much more by working with these agents. This is another step change in that progression."

Sources : CNBC

Published On : Feb 24, 2026, 19:10

Cybersecurity
Falling Debris from Aerial Interception Damages Oracle's Dubai Office Amid Rising Tensions

In a significant incident linked to escalating regional tensions, the Dubai office of Oracle sustained damage from debri...

CNBC | Apr 04, 2026, 09:25
Falling Debris from Aerial Interception Damages Oracle's Dubai Office Amid Rising Tensions
Science
Artemis II Mission Thrives as Astronauts Connect with Earth from Space

As the Artemis II mission entered its third day, the spacecraft's powerful engine had propelled the astronauts into a fa...

Ars Technica | Apr 03, 2026, 22:25
Artemis II Mission Thrives as Astronauts Connect with Earth from Space
Computing
Apple Navigates Its AI Challenge Amidst Celebrations and Setbacks

CUPERTINO, Calif. — In a momentous celebration at Apple Park, Nasdaq kicked off its market open festivities on the eve o...

CNBC | Apr 04, 2026, 12:20
Apple Navigates Its AI Challenge Amidst Celebrations and Setbacks
AI
The Alarming Rise of Cognitive Surrender: Are We Trusting AI Too Much?

Recent findings reveal a troubling trend among users of large language models (LLMs): a significant portion appears will...

Ars Technica | Apr 03, 2026, 21:10
The Alarming Rise of Cognitive Surrender: Are We Trusting AI Too Much?
Science
Ancient Dice Unearth Insights into Native American Understanding of Probability

A groundbreaking study reveals that Native Americans have been engaging in games of chance using dice for over 12,000 ye...

Ars Technica | Apr 03, 2026, 23:00
Ancient Dice Unearth Insights into Native American Understanding of Probability
View All News