Anthropic rebuilds Claude Code Projects for parallel agents

Anthropic has moved Claude Code further from a single assistant that answers one prompt at a time. The rebuilt Projects feature now has a coordinator split a stated goal across parallel threads, each running as its own cloud session that can open pull requests and run tests. A shared memory builds across those threads over time, and a library collects uploaded files and results.

That design fits alongside Anthropic's recent decision to make autopilot mode the default in Claude Code, where the company claimed it outperformed human developers on safety tasks. The company's pitch is that more autonomous agents likely mean more burned tokens, and control over how many tokens get burned shifts from users to the company selling them — like letting gas companies control the throttle in your car.

The beta's reach is the immediate test: it is limited to select Pro and Max subscribers using cloud sessions, with Team and Enterprise access later and local execution coming soon. Whether the parallel setup actually delivers on a full goal, rather than just running more sessions at once, is likely to determine how quickly that wider access arrives.