Anthropic redesigns Claude projects, letting users describe work in one conversation and have Claude manage it across parallel threads, starting in Claude Code
A new experience for Claude projects, now available in beta in Claude Code — https://claude.com/blog/projects- redesigned
Claude
Context & Ripple Effects
Claude Projects began in 2024 as a shared space for chats, documents and code with a 200K-token context window. Anthropic then made Claude Code more suited to concurrent work through its desktop multi-session interface, while Slack and messaging integrations extended where a coding session could be reached.
The redesigned Projects experience turns that collection of sessions into a coordinator-led workflow: a user can set out the work in one conversation while Claude allocates it across parallel threads. The rollout begins as a beta in Claude Code, concentrating the change on development work rather than the broader Claude product.
First-order effects
Claude Code beta users can manage multi-part development work from one persistent project conversation instead of manually coordinating separate sessions.
Anthropic makes Claude Code’s existing multi-session design a delegated workflow layer, with project context serving as the shared control point.
Second-order effects
Development teams evaluating AI coding tools will compare them less on a single chat’s output and more on whether one workspace can retain context and coordinate parallel tasks.
Cursor’s comparable long-running, shared-context approach, noted by a commentator, puts pressure on coding-assistant vendors to make multi-agent coordination understandable and controllable rather than merely available.
Third-order effects
If coordination from a single conversation becomes the standard interface, AI coding products may consolidate chat, task tracking, persistent project memory and execution management into one agentic work surface.
The durable competitive question shifts from how many agent threads a product can launch to how reliably it preserves context and gives users oversight of delegated work.
The trend: AI coding assistants are evolving from individual coding chats into persistent project workspaces that coordinate parallel agent activity.
Projects now run from one conversation, starting in Claude Code. You describe what needs doing, and Claude directs parallel threads that keep working after you close your laptop. In beta today for select Pro and Max users in cloud sessions; coming to all Claude users soon.
Been a minute but really proud of this one! Brought a bunch of what we learned from building Chat/Cowork projects to build something really powerful. In particular, Artifacts inside of Projects are incredible. Excited to see what people think!
Sounds similar to what Cursor announced recently, right? Long-running projects, with context and memories shared across cloud threads as one layer. Very intriguing! Also: I want this for regular Claude/Cowork too.
Super excited to show the world Projects!! We've been building this for the last few months and it has become my new daily driver. A few tips for getting the most out of it 🧵
We've been using projects for a while internally and this is how I do all my work now. Instead of needing to manage and check individuals sessions, you work with one higher level coordinator Claude that directs work for you.
my first biiiiiig launch at anthropic is finally here!! 🫶 projects are a huge labor of love from the claude code team - it's the only way i've been working with claude for the last several months (including using projects...to build projects!!) spam the coordinator with any and a…
Projects brings the architecture of Claude Tag to Claude Code. It has one agent per project that manages memory and spins off subagents for tasks. You can ask it to be proactive, to do things on a schedule, etc. It feels a lot nicer than a bunch of sessions, try it out!
Another day, another improvement to how we work with Claude, that makes it a lot more intuitive, starting with Claude Code: Projects are getting a big upgrade. …