Recently I’ve been spending more time in Claude Code on Desktop than in the CLI.
I’m finding that I can use the Desktop GUI to better manage multiple parallel work items, and then dive in with VS Code and/or Claude Code CLI for more surgical changes. I’ve never been a terminal poweruser or anything, so for me I find the Desktop GUI very useful for its status indicators, the ability to quickly flick between sessions, and the nice PR and CI integrations. I’m sure you can hack many of these things into your terminal but I’d rather leave that task to Anthropic’s product team personally!
My hunch is that IDEs of the future will be centered much more around an abstracted and/or conversational interface than they are today - with just the tiny little caveat that this hunch is predicated on coding agents demonstrating long term return on investment for enough of the industry. So with that hunch in mind, I’m taking the time to figure out what works (and what doesn’t) for me in this kind of interface, and so far I’m settling in quite nicely with Claude Code on Desktop.