Anthropic has expanded the Claude developer platform with Claude Managed Agents, a capability that lets teams deploy autonomous, stateful agents without managing their own infrastructure. According to Anthropic's release notes, the feature set includes dreaming — allowing agents to process tasks asynchronously — along with multiagent orchestration, webhook support, and a new outcomes layer for tracking agent results end-to-end. The launch arrived alongside higher API rate limits for Claude Code and Opus. Anthropic also shipped a Swift package enabling Apple developers to route complex queries from Apple's Foundation Models framework directly to Claude. The updates position Claude as an increasingly complete platform for production agentic applications, rather than just a model accessed through an API.