Anthropic Discloses AI Built Majority of New Claude Cowork Tool
SAN FRANCISCO — Anthropic used “mostly AI” to build its new Claude Cowork tool, the company disclosed, according to Mashable.
The AI safety company’s acknowledgment that Claude itself played a central role in constructing the Cowork feature reflects a broader shift in the software industry, where AI-assisted coding tools are moving from novelty to standard practice — even within the companies that build them.
Claude Cowork is designed to allow users to collaborate with Anthropic’s Claude AI model on complex, multi-step tasks, extending the assistant’s capabilities beyond single-turn conversations into sustained project work. The tool represents Anthropic’s push into the agentic AI space, where rival offerings from OpenAI, Google and others are vying for developer and enterprise adoption.
Anthropic’s disclosure serves as a demonstration of Claude’s coding capabilities for potential enterprise customers evaluating AI development tools.
The practice of AI companies “dogfooding” their own models — using them internally before or alongside external release — has become increasingly common across the industry. Google has previously disclosed using its Gemini models in internal development, and OpenAI has similarly described using GPT models in its own workflows.
Anthropic, which has positioned itself as a safety-focused alternative in the AI race, has raised billions in venture funding and counts Amazon as a major investor. The company continues to expand its product lineup as competition in the foundation model market grows.
The announcement comes as AI-assisted coding tools have seen rapid adoption in the enterprise software market, with GitHub Copilot, Cursor and a growing roster of competitors reporting growth among developers.
