Anthropic just launched a web app for Claude Code, letting developers spin up AI coding agents directly from their browser. The move breaks Claude Code free from its command-line roots, making the company's $500 million coding tool accessible across desktop and mobile platforms for the first time.
Anthropic is betting big on browser accessibility. The company launched Claude Code's web app Monday, marking a crucial pivot from the terminal-only tool that's become a developer favorite. Pro plan subscribers at $20 monthly and Max users paying $100-200 can now access Claude Code at claude.ai through a dedicated "Code" tab.
The timing couldn't be more strategic. AI coding tools are exploding into a crowded battlefield where Microsoft's GitHub Copilot once ruled unchallenged. Now Cursor, Google, and OpenAI are all pushing sophisticated coding assistants, many already web-native. But Claude Code's numbers speak volumes about its staying power - user growth jumped 10x since its May expansion, and the product now drives more than $500 million in annualized revenue for Anthropic.
"We're continuing to put Claude Code everywhere, helping it meet developers wherever they are," Anthropic Product Manager Cat Wu told TechCrunch in an interview. "Web and mobile are a big step in this direction." Wu attributes Claude Code's success partly to Anthropic's increasingly popular AI models, which have become developer darlings in recent years.
The shift represents more than platform expansion - it's about fundamentally changing how developers work. Unlike early AI coding tools that functioned like glorified autocomplete, Claude Code's agentic approach lets developers spin up autonomous coding agents. This evolution has transformed millions of software engineers into managers of AI assistants rather than hands-on coders.
Wu herself exemplifies this transformation. The former engineer says she "rarely ever sits down at a keyboard to write code anymore" and mostly reviews Claude Code's outputs. Anthropic claims 90% of Claude Code's own codebase comes from AI models - a meta-achievement that CEO Dario Amodei believes will soon become industry standard. He predicted earlier this year that AI should write 90% of code for software engineers.
But the transition isn't universally smooth. Recent studies show some engineers actually slow down when using AI coding tools like Cursor. Researchers point to time lost in prompting and waiting for AI responses, plus struggles with large, complex codebases where AI models often generate incorrect solutions. These growing pains haven't deterred Anthropic from pushing deeper into agentic coding.
The competitive landscape keeps intensifying. Cursor already launched its web app earlier this year, while GitHub Copilot maintains its integration advantage across Microsoft's developer ecosystem. Google and OpenAI are pushing their own coding assistants with increasing sophistication.
For Anthropic, the web launch isn't about abandoning the terminal - Wu emphasizes CLI will remain "the most intelligent and customizable way" to use coding agents. Instead, it's about meeting developers where they're already working, whether that's in browsers, mobile apps, or traditional development environments.
The company's approach includes what Wu calls "sprinkling in some fun" - a deliberate strategy to differentiate Claude Code in an increasingly crowded market. With $500 million in annual revenue already flowing from the coding tool, Anthropic has solid foundation to experiment with user experience and platform expansion.
Anthropic's web launch for Claude Code signals the maturation of AI coding tools from niche terminal applications to mainstream developer platforms. With $500 million in annual revenue and 10x user growth, the company's betting that accessibility will drive the next phase of adoption. But as the AI coding market grows more competitive and studies question productivity gains, success will depend on whether developers actually want AI agents managing their code - or if they prefer staying in control of the keyboard.
