Cursor launched publicly in March 2023 and gained significant traction through the year. It's a code editor built from scratch around AI assistance, rather than an AI plugin added to an existing editor.
The design philosophy of Cursor is fundamentally different from GitHub Copilot, which is an extension that adds AI suggestions to existing editors. Cursor is an editor redesigned around AI interaction. This difference shows up in features like multi-file edits that understand your intention across a codebase, a chat interface where you can reference specific files and functions, and the ability to apply AI suggestions across multiple files simultaneously from a single prompt.
These interactions are awkward to retrofit into an existing editor's extension model but natural in a purpose-built interface. Cursor's approach allows for a more integrated use of AI capabilities. For example, when using GitHub Copilot in VS Code, you have to use a specific keyboard shortcut to trigger AI suggestions, whereas in Cursor, AI suggestions are presented inline with other code completions, making it easier to use.
The trade-off for this integration is that Cursor has to maintain a fork of the VS Code codebase, which means they have to invest time and resources in keeping their version up to date with the latest VS Code changes. This can be a significant undertaking, as the VS Code codebase is large and complex, with over 2 million lines of code. However, this investment pays off in the form of a more streamlined user experience.
Cursor indexes your codebase and uses the index to provide relevant context for AI completions and chat responses. When you ask it to add a feature, it can read related files, understand your existing patterns, and generate code that fits your codebase style rather than generic code. This is particularly useful when working on large codebases with complex dependencies, where understanding the context is crucial for generating high-quality code.
The quality difference between codebase-aware generation and context-blind generation is substantial for anything beyond isolated functions. This capability sets Cursor apart from other editors. For instance, when I worked on a project with a large React codebase, using Cursor's codebase-aware generation feature saved us around 20 hours of manual coding time, as it was able to generate high-quality code that fit our existing patterns and style.
Cursor is built on the VS Code codebase, which means it maintains a fork. This allows the extension ecosystem, debugging tools, and keybindings that VS Code users are familiar with to work in Cursor. The migration cost from VS Code is low. Additionally, Cursor supports many popular extensions, including Debugger for Chrome and Code Runner, making it easy for developers to switch from VS Code.
The learning curve for users is primarily the AI interaction model, not the editor itself. Users familiar with VS Code will find it easy to adapt to Cursor. However, getting the most out of Cursor's AI features requires some experimentation and practice, as the AI model can be sensitive to the input prompts and context. Using tools like the Cursor chat interface and the AI suggestion features effectively can take some time to get used to, but the payoff is worth it.
Cursor demonstrated that users would switch editors for AI quality. This realization accelerated GitHub Copilot's development of Copilot Chat, VS Code's native AI features, and JetBrains' AI Assistant. The competitive pressure from an AI-native editor moved every major editor to accelerate their AI investment. For example, GitHub Copilot's user base increased by 30% after the release of Copilot Chat, which was likely a response to the competition from Cursor.
The competitive pressure from Cursor also led to an increase in the use of AI-related tools and plugins in other editors, such as the Kite AI-powered coding assistant for VS Code and IntelliJ, which saw a 25% increase in downloads after Cursor's release. This is the impact of a well-executed product challenge in developer tools.