"‘AI Programming Star’ Cursor CEO: The ‘Third Era’ of AI Software Development Has Arrived"

```

AI-assisted programming is undergoing a fundamental paradigm shift.

Recently, Cursor co-founder Michael Truell posted on X, stating that the company has officially entered the "third era" of AI programming. The core of this new era is driven by cloud agents that can independently handle complex tasks and operate autonomously over extended periods. This means Cursor’s positioning has fundamentally changed, evolving from a “tool for writing code” into a “platform that helps developers build software factories.”

The data is supporting this judgment: currently, 35% of merged code submissions (PRs) within Cursor are completed by autonomous agents running on cloud virtual machines. More notably, the company expects that within a year, the majority of software development tasks will be taken on by such agents. This trend will not only reshape the competitive landscape for AI programming tools, but also profoundly affect the business models of the entire industry.

From Tab to Agents: Rapid Reversal in User Behavior

The evolution of AI programming is surpassing everyone’s expectations. In the article, he recaps the key turning points of this field, clearly dividing the development of AI programming tools into three stages:

Stage One: The Tab Auto-Completion Era. Tab not only completes the current line but can intelligently predict and finish the next line of code, and even make changes across multiple files, liberating developers from tedious coding. This stage lasted nearly two years, with the core being automation of low-entropy, repetitive work.

Stage Two: The Synchronous Agent Era. The core function of this era is conversational programming—developers describe requirements to agents in natural language, and agents generate and respond with code in real time, forming a quick “prompt-feedback-revision” interaction cycle. He predicts this stage may last less than a year, with the speed of transition far exceeding previous expectations.

Stage Three: The Cloud Agent Era. After the developer hands over the task, the agent runs independently on a cloud virtual machine—completing code writing, debugging, testing, and iteration autonomously. Developers shift from "code writers" to "directors of agents."

User behavior data confirms the intensity of this paradigm shift: in March 2025, Cursor’s Tab users outnumbered agent users by 2.5 times; now, the ratio is fully reversed—the number of agent users is twice that of Tab users, and usage is still surging. He revealed that many Cursor users no longer use the Tab key at all.

Core Advantages of Cloud Agents: Parallelism and Asynchrony

Synchronous agents are limited by a dual constraint: real-time interactions with developers, and competition with local machines for computing resources. This greatly limits the number of synchronous agents that can run simultaneously.

Cloud agents fundamentally remove these constraints. Each agent runs on an independent cloud virtual machine. Developers can hand over tasks and switch to other matters without waiting in real time. Over hours, the agent autonomously completes code iteration and validation, finally delivering results in the form of logs, recordings, and real-time previews, instead of showing code differences line by line.

This delivery mode makes running multiple agents in parallel a reality. Developers can quickly assess the output quality of multiple tasks without rebuilding the context from scratch for each session. The human role is fundamentally transformed: from “line-by-line guiding code” to “defining problems and setting review standards.”

Internal Practice Reveals a New Developer Working Model

Cursor uses its own internal practice to illustrate the specifics of this new work mode. Developers adopting the new approach share three traits: agents write nearly 100% of the code; developers focus their time on breaking down problems, reviewing artifacts and code, and providing feedback; and multiple agents are launched in parallel rather than guided one-by-one to completion.

He also admitted, this model still faces challenges for large-scale industry adoption. On an industrial scale, unstable tests or damaged environments that a single developer can sidestep may evolve into systemic failures that interrupt every agent run. Furthermore, ensuring agents have access to all necessary tools and context remains a critical issue to be solved.

Meanwhile, he said that Cursor's recent feature releases are a “preliminary but important step” in this direction.

Risk Disclaimer and Exemption ClauseThe market carries risks, and investment requires caution. This article does not constitute personal investment advice and does not take into account individual users’ specific investment objectives, financial conditions, or needs. Users should consider whether any opinions, viewpoints, or conclusions in this article are suitable for their particular situation. Investment is at the user's own risk. ```