Anthropic appears to be moving Claude Code beyond the single-agent workflow. A reported relaunch of Projects would let developers organize several Claude Code agents around one assignment, with shared goals, memory, files and artifacts instead of treating every coding session as a separate conversation.
Each Project can contain multiple threads working in parallel, while a coordinator keeps the broader assignment organized. The feature is reportedly entering beta for select Claude Pro and Max subscribers, with wider access planned for Pro, Max, Team and Enterprise users.
Claude Code Is Becoming A Workflow, Not Just A Coding Assistant
The important shift is structural. Each thread reportedly runs as its own Claude Code cloud session, with a separate branch and copy of the repository. That gives individual agents room to handle different tasks without forcing every decision through one long chat. Users can work with a thread directly or monitor the whole assignment from the main Project conversation.
The coordinator is the useful part. It gives the work a management layer, dividing a large brief into parallel tasks and keeping the outputs connected. Threads can also split their own assignments into subagents, loops and workflows. In theory, a complex software job can move from one assistant answering questions to a small team handling research, implementation, testing and review at the same time.
The Hard Part Is Still Coordination
Anthropic’s reported design also exposes the friction that comes with agent teams. If two threads change the same code, the overlap becomes a merge conflict, just as it would in a conventional pull request. More agents do not remove the need for decisions about ownership, review and quality. They make those decisions more important.
The feature reportedly starts with cloud-based work, while support for local tools and code is expected soon. That limitation matters because serious development work often depends on private systems, custom environments and tools that cannot simply be moved into a hosted session.
Still, the direction is clear. Claude Code is being shaped less like a chatbot for developers and more like a workspace where people supervise a group of specialized agents. The competitive question will not only be which model writes the best code. It will be which product gives teams the clearest way to divide work, inspect decisions and resolve the mistakes created by parallel speed.