Anthropic introduces "dreaming," a system that lets AI agents learn from their own mistakes

Anthropic added “dreaming” to Claude Managed Agents, a scheduled memory-and-review feature that helps agents learn from past sessions.

The company also moved outcomes and multi-agent orchestration into public beta, expanding capabilities for enterprise workflows.

Early users reported better task completion, faster document review, and more parallel build processing.

Anthropic said revenue and usage in early 2026 are growing well above internal expectations, signaling strong demand for its platform.