Anthropic’s new Claude Code feature could drain your plan before lunch
The New Stack Amanda Caswell ● Covered by 3 sources
Anthropic is letting Claude Code run multiple jobs at once. That means faster coding work, and faster plan burn if you’re not watching.
Based on reporting by The New Stack, Amanda Caswell — read the original for the full story.
Summary, retelling and take written by AI under human oversight; images are AI-generated illustrations. How we work · Report an error
Anthropic is turning Claude Code into a coordinator, not just a coding assistant. Starting Thursday, select Claude Pro and Max subscribers can try a redesigned Projects beta through cloud sessions in Claude Code, with access widening to more users on those plans over the next week.
The old Projects setup mostly grouped chats around a shared knowledge base and instructions. The new version adds a central organizer that can take an engineering goal, split it into smaller tasks, and send them to multiple Claude Code sessions running in parallel. Anthropic says that cuts out the manual juggling developers used to do when they had to split work, run it in separate sessions, and stitch the results back together.
The company’s own description is blunt: Claude scopes the request, delegates the work, coordinates parallel threads, reviews the outputs, and assembles the finished result. Developers set the goal, Claude decides how to break it up, and each thread can be watched from the main Project chat or opened on its own if someone needs to step in. Every thread is its own cloud session, with a separate branch and its own copy of the repository. Those worker threads can also use subagents, loops, and workflows to chip away at smaller pieces.
There’s a catch, and it’s not a subtle one. Anthropic says Projects will hit usage limits sooner when several threads are active, because each thread counts as a full Claude Code session. That matters now because Anthropic is already dealing with a class-action lawsuit from Max plan subscribers who say promised usage increases came with weekly ceilings the company did not disclose at first.
Anthropic is also leaning harder into coordination and memory. Shared context can carry details across threads, including a changed release date, why a feature was dropped, or who has to approve a service. The system can also remember how a project should be run, from check-in frequency to how detailed updates should be. Team and Enterprise access is planned for later, mobile support is coming soon, and subscribers who already use Projects stay on the current version until they’re upgraded.
My take — AI-written commentary, not fact-checked reporting
This is the kind of feature that makes perfect sense to product people and perfect pain to anyone paying by usage. Anthropic is selling orchestration, but orchestration is just a nicer word for “more sessions, more bill, please.” The real test isn’t whether Claude can juggle threads; it’s whether Anthropic can stop turning every productivity win into a quota trap.
Read more about this at: The New Stack