Anthropic’s "Code with Claude" developer event introduced significant updates to Claude Code and the Claude API, focusing on enhancing agentic capabilities and developer workflows. Key features include "Routines," which enable scheduled or webhook-triggered tasks, and "Outcomes," a system allowing agents to iterate against a defined rubric until a task meets success criteria. The event also unveiled a multi-agent framework supporting up to 25 coordinated agents with distinct toolsets, and "Dreams," a research-preview feature for consolidating agent memory across sessions. Additionally, Anthropic increased usage limits for Claude Code and API models, expanding capacity for developers. These updates prioritize practical, agent-driven workflows, signaling Anthropic’s strategic push to become the primary platform for building sophisticated AI agents.
Sign in to continue reading, translating and more.
Continue
