Cursor just dropped 2.0 and I’m telling you—this isn’t a “nice update”, it’s a complete rethink of how we write code. In this video I walk through what actually changed the moment you open Cursor 2.0: the new Agents-first UI, the new Plan option, browser connect, and voice input. Then I go deep on the real headline feature: Composer, Cursor’s in-house coding model. It’s built for software engineering workflows, understands your codebase conventions, and in my demo it took a task from 27 seconds on my usual Claude Sonnet 4.5 reasoning flow down to ~10 seconds on Composer. That speed + project awareness combo is genuinely game-changing. Then I show why developers are getting obsessed: the in-app browser (with hot reload + DevTools) so the agent can test while it codes, and multi-agent parallelization where up to eight agents work in isolated sandboxes (separate git work trees) and you compare diffs to pick the best solution. I also cover voice mode (surprisingly useful, occasionally hilarious) and Plan mode, which I love as a senior dev because it separates architecture from execution—plan first, review assumptions, then build. My takeaway: you shouldn’t be coding line-by-line anymore; you focus on architecture and product decisions, and let Cursor handle implementation.