/
Navigation
Chronicles
Browse all articles
Explore
Semantic exploration
Research
Entity momentum
Nexus
Correlations & relationships
Story Arc
Topic evolution
Drift Map
Semantic trajectory animation
Posts
Analysis & commentary
Pulse API
Tech news intelligence API
Browse
Entities
Companies, people, products, technologies
Domains
Browse by publication source
Handles
Browse by social media handle
Detection
Concept Search
Semantic similarity search
High Impact Stories
Top coverage by position
Sentiment Analysis
Positive/negative coverage
Anomaly Detection
Unusual coverage patterns
Analysis
Rivalry Report
Compare two entities head-to-head
Semantic Pivots
Narrative discontinuities
Crisis Response
Event recovery patterns
Connected
Search: /
Command: ⌘K
Embeddings: large
TEXXR

Chronicles

The story behind the story

days · browse · Enter similar · o open

Anthropic updates Claude Managed Agents with “dreaming”, a scheduled process that reviews recent sessions and updates Claude's memory, in research preview

The New Stack Frederic Lardinois

Context & Ripple Effects

Anthropic’s recent agent work has moved from coordinating multiple Claude Code agents in parallel to allowing work to continue in the background. Related coverage had already indicated that a “dream mode” was intended to consolidate memories between work periods.

This preview brings that memory-consolidation concept to Claude Managed Agents. It matters because persistent context can make agents more useful across repeated sessions, rather than treating each task as isolated.

First-order effects

  • Claude Managed Agents in the research preview can review recent sessions on a schedule and update their memory, giving Anthropic a mechanism for continuity across work cycles.
  • Teams testing Managed Agents gain a new way to preserve relevant prior-work context without relying solely on users to restate it in later sessions.

Second-order effects

  • The feature makes Anthropic’s parallel-agent and background-work capabilities more complementary: agents can coordinate on work that persists beyond a single active session.
  • Competing coding and enterprise-agent products face greater pressure to pair autonomous task execution with controllable long-term memory, not just larger context windows.

Third-order effects

  • If managed memory proves reliable, agent platforms may increasingly be judged as persistent work systems—combining delegation, background execution, and accumulated context—rather than as turn-by-turn chat interfaces.
  • The same persistence raises a durable product-governance challenge: providers will need clear controls over what an agent retains, updates, and carries into later work.

The trend: AI vendors are assembling persistent, multi-agent work environments in which memory and asynchronous execution are core capabilities rather than add-ons.

Discussion

  • @unusual_whales @unusual_whales on x
    BREAKING: Anthropic has released a feature called “dreaming” which allows AI agents to self-improve
  • @claudedevs @claudedevs on x
    In Claude Managed Agents, we've added multiagent orchestration, an outcomes loop for rubric-driven self-improvement, dreaming for self-learning, & webhooks. [video]
  • @claudedevs @claudedevs on x
    Agents write to memory in a session. Dreaming is a background process that reflects over many sessions to curate memories: it can edit them based on patterns, add new skills, or remove stale ones. https://platform.claude.com/ ...