Cursor 3 announced with AI coding agents to automate developer tasks, what will humans do?

The way developers write code has been going through quiet changes for a while, but the latest announcement from Cursor makes it hard to ignore what is coming next. What used to be a simple "AI helping you code" story is now turning into something else entirely - AI doing the work on your behalf.

Cursor has introduced Cursor 3, a new interface that lets users assign coding work to AI agents instead of doing it themselves. The product, which was earlier being developed under the name "Glass," allows developers to describe what they want in plain English, and the system handles the rest.

At the centre of this experience is a chat-like box. You type a task, hit enter, and the AI agent starts working. There is no need to open files, write functions, or debug line by line, at least not at the starting point.

"In the last few months, our profession has completely changed," Jonas Nelle, one of Cursor's heads of engineering, told WIRED. "A lot of the product that got Cursor here is not as important going forward anymore." This is a bold admission, and it tells you how quickly things are moving.

Cursor isn’t alone in trying this approach. Companies like OpenAI and Anthropic are already offering similar tools like Codex and Claude Code that let developers hand over entire chunks of work to AI.

These tools have caught on fast. Many developers have started spending more time with them, especially because of how aggressively they are priced. Some subscriptions reportedly offer usage worth over $1,000 for a monthly fee of around $200, making them hard to ignore. That has likely created pressure on Cursor. Some developers who once relied on it are now trying other platforms, mainly because they feel they are getting more value for money elsewhere.

How Cursor 3 actually works in real life

Cursor 3 doesn't remove the old way of coding completely. Instead, it adds a new layer on top of it. Inside the app, human developers can run multiple AI agents at the same time. Each agent can be assigned a different task - building a feature, fixing a bug, or writing a chunk of code. While these agents work in the background, users can keep track of them through a sidebar.

One interesting part is how Cursor connects this system with its existing development environment. For example, you can ask an agent to create a feature in the cloud and then review the generated code on your own machine. So while the AI does the heavy lifting, developers still get a chance to check and refine things.

So, what happens to human developers now? What will they do?

This is the question that naturally comes up. If AI can handle entire tasks, where does that leave humans? Right now, developers are still very much in the loop. Someone has to decide what needs to be built, give clear instructions, and verify that the output actually works. AI can move fast, but it doesn’t always get things right.

What is changing is the day-to-day experience. Instead of spending hours writing and fixing code, developers may spend more time reviewing, tweaking, and managing multiple AI agents. Some may welcome this. Others may find it uncomfortable, especially those who enjoy the craft of coding itself.