AI Briefing
KO

Anthropic Developing Claude-based Managed Projects

·2026.07.23 09:00

Key point

Anthropic is internally testing a 'Managed Projects' feature in which Claude autonomously performs tasks and manages projects.

1 / 2

Details

Anthropic is developing a new project type called Managed Projects. When creating a project, users can choose between the existing standard project type and a managed project, and in managed projects, Claude directly performs tasks and organizes and maintains the project. Going beyond simply saving files and answering questions, it aims to be a persistent environment capable of executing scheduled tasks and autonomously maintaining context.

This feature is in line with several technologies Anthropic has already released. Claude Managed Agents, available on the developer platform since April, maintains state across sessions and refines memory through a scheduled process called 'dreaming.' Claude Code also maintains context across sessions through project memory and instructions. Managed Projects can be seen as a form that integrates these features into a general user interface.

Projects are divided into personal and shared types, and shared projects will be offered under the Team and Enterprise plans. Meanwhile, there is speculation that this could signal a transition, since Anthropic's always-on internal agent Conway is scheduled to end on July 24. Currently, this is in an internal testing stage available only to employees and a small number of trusted testers, and there is no set timeline for a full release.

This summary was generated automatically by AI. Check the original for the author's claims and context. Copyright belongs to the original author.

Our guide explains how the AI works. Report summary errors, attribution issues, or removal requests via Contact.