If it's been a few months since you last poked around Claude's interface, a handful of things have shipped that are worth knowing about - not because you need to use all of them, but because a couple genuinely change how you'd approach certain tasks.
As a refresher: Claude now spans three surfaces - the chat window, Claude Code, and Cowork - covered in more depth in The Six Decisions. What's new sits mostly inside and around those, not replacing them.
Chat: small features that solve real annoyances
A few additions to regular chat conversations are easy to miss but genuinely useful once you know they exist:
- Fork. Branch a conversation at any point instead of editing a message and losing the original thread - handy when you want to try two different directions from the same context.
- /rewind. Jump back to an earlier point in a long conversation without scrolling through everything to find it.
- /recap. Get a quick summary of a long conversation so far, useful when you're picking a thread back up after a break.
- /btw. Drop a side note into the conversation without derailing the main thread Claude is working on.
- Output Styles. Set a default tone or format for responses so you're not re-explaining "keep it brief" or "use a formal tone" in every prompt.
- Inline interactive charts. Charts render directly in the conversation and can be interacted with, instead of static images you have to describe follow-up changes to.
None of these require learning anything complicated - they're worth trying the next time you're already deep in a long or messy conversation.
Claude Design: the bigger shift
The one genuinely new surface is Claude Design, a prototyping workspace built for turning descriptions into working visual drafts - UI mockups, slides, one-pagers, that kind of thing.
How it works, in practice: you get a two-pane layout with chat on the left and a live canvas on the right. You describe what you want, Claude generates it on the canvas, and you refine it through comments, voice, or direct adjustment controls for things like spacing and color - closer to working with a designer than prompting a chatbot.
The detail that matters most for teams: during onboarding, Claude Design can read your existing codebase and learn your actual colors, typography, and components, so what it generates afterward matches your real design system instead of generic defaults. It also round-trips with Claude Code, so a finished design can be handed off to build the real interface without starting over - or you can pull up a /design command right inside a Claude Code session.
Does this actually matter for you?
If you're a designer, PM, or founder who regularly needs quick mockups or prototypes, Claude Design is worth a real look - it's a meaningfully different workflow from describing a design in words and hoping the output matches what you pictured. If your work is mostly writing, analysis, or automation rather than visual UI, you can safely skip it and focus on the smaller chat features above, which pay off in almost every conversation.
/recap to see the summary Claude generates - it's the fastest way to feel the difference these updates make.