Anthropic Introduces Redesigned Projects in Claude Code

Overview of the Redesigned Projects Experience

Anthropic has officially announced a significant update to how users interact with Projects in Claude Code. By shifting from a folder-based management style to a conversation-centered architecture, the platform aims to streamline the way users handle complex, multi-part development tasks. This redesign is currently available in beta for a select group of Claude Pro and Max subscribers who utilize cloud sessions within the Claude Code environment.

Historically, managing a technical project required manual intervention, including dividing work into smaller chunks, navigating handoffs, and manually reassembling components. The updated Projects feature changes this workflow by introducing a coordinator model. Users define a primary goal, and Claude assumes the role of an orchestrator that scopes the request, delegates specific tasks to parallel threads, reviews the results, and ensures the final objectives are met.

How Agentic Threads Operate

The core of this new system lies in its use of parallel threads. Each thread serves as a dedicated, cloud-based Claude Code session. These sessions are autonomous, meaning they can continue executing tasks—such as profiling endpoints, running tests, or opening pull requests—even after a user disconnects from the session or closes their computer. When a user provides a brief for the project, Claude functions much like a chief of staff, directing work to both new and existing threads depending on the context.

Because these threads operate within their own isolated branches and repository copies, they allow for high concurrency. If multiple threads interact with the same section of a codebase, the system handles potential overlaps by managing them as standard merge conflicts. This structure is specifically optimized for long-running workflows that typically exceed the scope of a single interaction or response cycle.

Enhanced Memory and Context Management

Beyond task execution, the redesigned Projects include a unified memory system. As work progresses, individual threads contribute to and draw from a shared pool of information. This reduces the necessity for repeated prompt engineering, as the system retains context regarding project-specific details, such as scheduling changes, team communication preferences, or technical constraints. Users also have the ability to guide the assistant's behavior, adjusting parameters like the frequency of status updates or the level of detail provided by the system.

To support this, a new library feature has been introduced to consolidate uploaded files and AI-generated artifacts. By centralizing these resources, Anthropic aims to make it easier for users to track historical decisions and build upon previous efforts within a project.

Availability and Future Rollout

As of September 17, 2026, the updated Projects are in a limited beta phase. The initial release is targeted at Claude Pro and Max subscribers who use cloud sessions and do not currently have existing projects configured on the web or desktop interfaces. Anthropic has stated that it plans to expand access to more users within the coming week. Following this initial testing period, the company intends to integrate these capabilities across the broader Claude ecosystem, including team and enterprise plans, as well as standard chat and the Cowork environment.

Users on existing Pro and Max plans can continue to use their current projects without interruption. Anthropic confirmed that these legacy projects will be upgraded as the rollout progresses to ensure a smooth transition to the new, thread-based architecture.