Cursor declares a third era of AI development, and backs it with data
Cursor has published a manifesto, "The Third Era of AI Software Development," describing three eras of how developers work with AI. The first was code autocompletion, the second chat assistants, and the third full agentic development, where AI solves tasks on its own. Cursor's internal statistics indicate the move to the third era is already under way. The share of agentic interaction is growing fast, and developers increasingly act as architects and reviewers rather than authors of every line of code.
AI-processed from Habr AI; edited by Hamidun News
When a company whose product hundreds of thousands of developers around the world use every day publishes a manifesto about a paradigm shift in programming, it deserves attention. Last week the Cursor engineering blog released a text titled "The Third Era of AI Software Development" — a short but dense document backed not by abstract reasoning but by real data on how developer behavior is changing right now.
Cursor is one of the most successful AI-oriented code editors, built on VS Code and deeply integrated with large language models. The company is in a unique position: it sees in real time how millions of engineers interact with AI while writing code. That vantage point makes its classification especially valuable.
According to the manifesto, the first era of AI development was the era of autocompletion. It began with GitHub Copilot and similar tools that suggested the continuation of a line or block of code based on context. The developer stayed fully in the driver's seat, and the AI acted as an advanced autofill — useful, but not fundamentally changing the process. The programmer thought and wrote, while the machine occasionally guessed the next few tokens.
The second era is chat assistants. ChatGPT, Claude and chats built into IDEs turned interaction with AI into a dialogue. A developer could ask a question, request an explanation of a code fragment, or generate a function from a description. It was a qualitative leap: AI became a conversation partner and co-author. But the initiative still rested with the human — every action required an explicit request, and every result had to be integrated into the project by hand.
The third era, whose arrival Cursor is announcing, is agentic development. Here AI stops being a tool that answers questions and becomes an autonomous agent able to explore a codebase on its own, plan changes, edit multiple files, run tests and iterate until it gets a working result. In this model the developer shifts from author of every line to architect, task-setter and reviewer.
The most important part of Cursor's publication is not the classification itself but the data behind it. The company's internal statistics show rapid growth in the share of agentic sessions in overall product usage. This is not a theoretical prediction or a marketing narrative — it is a measurable shift in the behavior of real engineers, who are voting with their working hours for a new way of interacting with code.
The trend matches what is happening across the industry. Anthropic is developing Claude's agentic capabilities, OpenAI is investing in Codex and agent frameworks, and Google DeepMind is experimenting with autonomous coding agents. The biggest players have converged on the same conclusion: the future of development belongs to systems that do not merely help write code but take on entire tasks, from specification to implementation.
The consequences of this transition reach well beyond the convenience of an individual programmer. If the third era is really arriving, it means a fundamental rethinking of what it means to be a developer. Skills that were considered key for decades — typing speed, knowing syntax by heart, holding complex data structures in your head — are giving way to other competencies: the ability to formulate tasks clearly, decompose problems, assess the quality of generated code and see the whole architectural picture. The economics of development are changing too: if one engineer with agentic tools can do work that once required a team, that will inevitably affect the structure of the labor market and which projects become economically viable.
Still, a healthy dose of skepticism is warranted. Cursor is a commercial company with an interest in promoting the narrative that its product is indispensable. Agentic development is still far from perfect: autonomous agents often go down dead ends, generate brittle code and require significant oversight. The third era may already have begun, but it is far from maturity. Even so, the direction of travel is clear — and those building a career in development should start adapting now.
- Nov 2, 2026Karpathy builds a YouTube clone with 6454 videos from Hacker News recommendations in two hours
- Nov 2, 2026Bun creator uses Claude Code to port 750K lines from Zig to Rust in 11 days
- Nov 2, 2026mcp-linter finds 18 undescribed parameters in Anthropic's filesystem server
- Nov 2, 2026The Verge reviewer builds a $50 Meta Muse Charm alternative with Claude Code
Build & Ship Production AI Agents Without Boilerplate Code
Start your free 72-hour trial at Hamidun Academy (academy.hamidun.com) — interactive lessons on Claude Code & Cursor, live Tech Radar, and production agent templates.
The AI world, distilled — once a week
Seven stories that actually mattered, hand-picked. No noise, no reposts, no press releases.
Done! Check your inbox for a confirmation.