Cursor 3.1 refines the agent-first workspace introduced in version 3. The parallel agent interface now handles handoff between cloud and local execution more smoothly. Bugbot resolution rate climbing to 78% after gaining self-improvement capabilities is a milestone for automated PR review.
The Pragmatic Engineer survey provides hard data on developer preferences - Claude Code leads at 46% most loved, while Cursor remains the most widely used AI coding environment. A New Stack analysis found the three major tools are converging into a composable stack rather than competing for winner-take-all.
Founder Takeaway
AI coding tools are becoming composable - build integrations with Cursor, Claude Code, and Codex rather than betting on a single platform.
