Skip to content
Breaking:

Anthropic Relaunches Projects in Claude Code to Support Multi-Agent AI Workflows

The updated feature lets developers orchestrate teams of Claude agents in cloud sessions with shared memory, file repositories, and merge conflict resolution.

By The Company Wire3 min read
Share
Anthropic — Anthropic Relaunches Projects in Claude Code to Support Multi-Agent AI Workflows
Anthropic — Anthropic Relaunches Projects in Claude Code to Support Multi-Agent AI Workflows. Photo: The Verge.

Anthropic has launched a revamped version of its Projects capability within Claude Code, enabling developers to orchestrate multiple artificial intelligence agents simultaneously, as first reported by The Verge. The redesigned feature is designed to allow distinct AI assistants to collaborate in the cloud on complex software engineering tasks.

The updated system unifies several AI agents under a single project framework where they share background context, operational objectives, and a centralized repository of files and digital artifacts. The setup resembles existing multi-agent management platforms such as Grok Bot, organizing complex development tasks into individual operational streams.

At the structural level, each project operates through parallel processing lines called threads, all managed by a central coordinator agent. The coordinator is responsible for assigning workloads, tracking overall progress, and maintaining organization across all active tasks within the project workspace.

Behind the scenes, every thread functions as an isolated Claude Code cloud session running on its own dedicated code branch and clone of the software repository. When multiple threads make adjustments to identical code files, the system manages overlapping edits by converting them into merge conflicts, mirroring standard developer pull request workflows.

To speed up execution on broad assignments, individual threads can break down their designated tasks into smaller sub-tasks. These components can then be delegated to lower-level subagents or run through specialized automated loops and tailored processing workflows.

Users are given flexible controls to manage these AI workflows. Engineers can communicate directly with specific task threads to give tailored instructions or use a primary project chat interface to oversee and adjust the broader collective project status.

For its initial release, all project execution threads are hosted and executed entirely in cloud environments. However, Anthropic indicated that functionality for local developer tooling and local codebase integration will be added in an upcoming update in the near future.

The redesigned Projects functionality is currently rolling out in a beta release limited to select Claude Pro and Claude Max subscription tiers. Anthropic plans to broaden access over time to include all Pro, Max, Team, and Enterprise account holders, along with integration into Cowork and standard Claude chat environments.

Sources

  1. The Verge

Company: Anthropic

Written by

The Company Wire

Newsroom · San Francisco

Inside the companies building what’s next. Reporting on startups, technology, funding and the people shaping them.