Anthropic's Claude Managed Agents can now "dream," sort of
Ars Technica • • technology
Key Points:
- Anthropic introduced “dreaming” for Claude Managed Agents, a process that reviews recent events to identify and store important memories for future tasks, currently available in research preview.
- Dreaming analyzes sessions and memory across multiple agents, surfacing patterns and important information that single-agent compaction processes typically miss, enhancing long-term and multi-agent projects.
- Users can opt for automatic dreaming or manually review memory changes, with the feature aimed at improving memory relevance and team workflow efficiency.
- Anthropic expanded access to other research preview features like outcomes and multi-agent orchestration and increased usage limits for Pro and Max subscribers due to high demand on its compute infrastructure.
