On February 24, 2025, Anthropic announced Claude Code as part of its post “Claude 3.7 Sonnet and Claude Code.” The page describes Claude Code as Anthropic’s first agentic coding tool and states that it was made available as a limited research preview, letting developers join to get access to the same tools Anthropic uses internally. This date and the research-preview framing come directly from the announcement; the page anchors the introduction rather than a general-availability date.
Claude Code differs from earlier code assistants in being agentic: rather than only suggesting completions inline, it is designed to take a goal and carry out multi-step work across a codebase, such as reading files, making changes, and running commands. Anthropic’s dedicated Claude Code page presents it as an AI coding agent that works in the terminal and integrated development environments.
For businesses, Claude Code reflects the shift from AI that autocompletes code to AI that can act as a hands-on collaborator on software tasks. It is part of the broader agentic turn in 2024 and 2025, in which AI systems move from answering questions to taking actions, with software development as one of the earliest and most measurable proving grounds.