Anthropic Launches Parallel AI Coding Agents in Claude

Anthropic has rebuilt its Claude Projects environment to deploy and manage multiple Claude Code cloud sessions simultaneously. Under this new architecture, a central coordinator agent receives a developer's high-level instructions, breaks the objective down into smaller tasks, and dispatches them to parallel worker threads. Each worker operates as an independent Claude Code session on its own Git repository branch, running tests and opening pull requests. Because these threads execute in the cloud, they continue running even after a developer closes their laptop, and can be monitored or redirected via a mobile device.

The system features a shared project memory that preserves context, ownership details, and past decisions across all active threads. This setup reduces the need to repeatedly feed background information into separate chats. However, running multiple parallel sessions will consume plan limits much faster, as each worker counts as a full Claude Code session. Users on the Pro and Max tiers can manage these costs by inspecting project-specific usage and adjusting the model and effort level for both the coordinator and worker threads.

For software practitioners, this update simplifies complex workflows like optimizing p75 latency or retiring a deprecated v1 endpoint across web, mobile, and API repositories. Instead of manually coordinating handoffs, developers can let Claude spin up separate threads for each repository, run tests, and outline the necessary merge order. Currently, the feature is in beta for select Claude Pro and Max subscribers who run cloud sessions and do not have existing projects. It remains cloud-only, meaning workers cannot access local files or private VPN networks, though Anthropic plans to add local resource support soon.