Anthropic's Claude Managed Agents Can Now "Dream," Sort Of
Also, 5-hour usage limits will double for Pro and Max users of Claude Code.
TL;DR
- Anthropic's Claude Managed Agents now feature "dreaming," a process for identifying and storing important information from past interactions.
- Dreaming helps overcome LLM context window limitations by curating and saving significant patterns across agents and sessions.
- This feature is in research preview and available to select developers; users can opt for automatic or manual memory review.
- Anthropic also made outcomes and multi-agent orchestration more widely available.
- Five-hour usage limits for Claude's Pro and Max subscription plans will be doubled due to user demand and infrastructure challenges.