Updates·September 17, 2026, 20:59
Claude Code can now run multiple AI agents at once
AI-generated and checked against the sources listed below.
Anthropic has updated its coding tool Claude Code so you can put several artificial intelligences to work on a programming task together. The feature is being tested with selected paying users.

Anthropic, the company behind the chatbot Claude, has made a new version of the Projects feature in its coding tool Claude Code. The tool is used by programmers to write and fix computer code with the help of artificial intelligence. What's new is that you can now put several AI agents, meaning artificial intelligences that can carry out tasks on their own without being constantly steered step by step, to work on a big task together.
Previously, users had to split a big task into smaller pieces themselves, start an agent on each piece and finally put the results together on their own. Now you can just describe what you want done, and the system takes care of the rest.
How the coordinator divides the work
It works by having a kind of controlling AI, called a coordinator, assess the task and distribute the work to several other AI agents. Each agent works on its own copy of the code and can, if necessary, start more helper agents itself. According to the outlet The Verge, it is resolved automatically if two agents collide in the same piece of code, meaning they make conflicting changes, much like when two human programmers work in the same file at the same time.
An important element is that all the agents in the same project share memory. This means they draw on the same goals, files and earlier decisions, so if, for example, a new deadline is decided in one part of the work, it can automatically affect the other agents' work too. According to Anthropic, this is meant to make things easier, so you don't have to start over or write long, detailed messages for every single task.
When the work is done, the coordinator gathers all the agents' contributions, and the result comes back to the user as an ordinary request to approve the changes, the same way a human colleague asks to have their code reviewed before it's put into use.
Rollout and price
The feature is still in testing and is rolling out in three stages. First, a smaller group of subscribers on Claude Code's Pro and Max plans get access if they already use the tool's cloud-based features. In the following week, access expands to more users on the same plans, and later the plan is for it to come to business customers too.
There's no extra charge for the feature. It's included in the usage you already have as a Pro or Max subscriber, where Pro typically costs about $20 a month and the Max plans start at about $100. You can sign up for a waitlist at Claude.com if you don't already have access.
It's not yet known whether and when the feature will come to Denmark and the rest of the EU, and there is no information about Danish language support. Another limitation is that the agents work only in the cloud and can't access files on the user's own computer or the company's internal network, but Anthropic has said that option is coming soon.
What it means for you
If you have a Pro or Max subscription to Claude Code, you can soon sign up for the waitlist and try the feature at no extra cost. You can now ask for an entire task to be solved at once instead of splitting it up yourself and steering several agents manually. Remember that the code still has to be approved by you before it's put into use. If you need to use your own files or your company's internal systems, you'll have to wait a little longer.
Sources
Get the week's AI news in your inbox
Choose your level, topics and length. One email a week, unsubscribe at any time.
Subscribe to Promptly Newsletter



