Anthropic Launches Dreaming in Claude Managed Agents, Enabling Between-Session Memory
Original: Anthropic Launches Dreaming in Claude Managed Agents, Enabling Between-Session Memory View original →
New Features Unveiled at Code with Claude 2026
On May 6, 2026, Anthropic announced new capabilities for Claude Managed Agents at its developer event, Code with Claude 2026. The headline feature, Dreaming, is now available as a research preview with a waitlist on Claude Console.
What Is Dreaming?
Dreaming is a scheduled process that reviews recent agent sessions and memory stores, extracts patterns, and automatically updates memory files. This enables agents to continuously learn from past interactions without human intervention—improving performance over time as they accumulate and apply relevant experience across sessions.
Additional Updates in Public Beta
Anthropic simultaneously moved three features to public beta: Outcomes (a rubric-based evaluation system for agent goals), multiagent orchestration, and webhooks. Together, these additions make Claude Managed Agents a more capable platform for building long-running, automated AI workflows directly from Claude Console.
Related Articles
Anthropic analyzed millions of real Claude interactions and found the 99.9th percentile session duration nearly doubled to 45+ minutes in 3 months, with software engineering accounting for nearly half of all agentic use.
Anthropic says Claude contains a J-space that resembles a global workspace for active, verbalizable thoughts. The lead tweet has more than 9.1 million views and points to audit use cases, including hidden goals in sabotage-trained models.
Anthropic released ten ready-to-run agent templates for financial services including pitchbook creation, KYC screening, and month-end close. Claude now works directly in Excel, PowerPoint, Word, and Outlook.