Anthropic geeft AI-agenten vermogen om van eigen fouten te leren

Anthropic has introduced "dreaming," a new capability for its Claude AI agents that enables them to learn from past sessions and improve over time. The company also moved two experimental features—outcomes and multi-agent orchestration—into public beta. Early adopters report significant improvements: legal AI company Harvey increased task completion rates by 6x, Wisedocs cut document review time by 50%, and Netflix now processes hundreds of simultaneous build logs. Anthropic disclosed experiencing 80x annualized growth in revenue and usage during Q1 2026, far exceeding its projected 10x