On Monday, Anthropic introduced a web-based application for its widely used AI coding assistant, Claude Code, enabling developers to launch and oversee multiple AI-powered coding agents directly from their browsers.
The web version of Claude Code is now being made available to users subscribed to Anthropic’s Pro plan at $20 per month, as well as those on the $100 and $200 per month Max plans. Both Pro and Max subscribers can use Claude Code online by visiting claude.ai—the same site as Anthropic’s consumer chatbot—and selecting the “Code” tab, or by accessing it through the Claude app on iOS.
This release represents Anthropic’s latest move to expand Claude Code beyond its original command-line interface, which developers previously accessed via terminal. By bringing Claude Code to the web, Anthropic aims to make it easier for developers to deploy AI coding agents in a wider range of environments.
The market for AI coding assistants is becoming more crowded as tech companies strive to differentiate their offerings. While GitHub Copilot from Microsoft once led the field, competitors like Cursor, Google, OpenAI, and Anthropic now offer advanced AI coding solutions—many already accessible online. Among these, Claude Code stands out as one of the most widely used. Since its broader release in May, the user base has grown tenfold, and the tool now contributes over $500 million annually to Anthropic’s revenue.
In an interview with TechCrunch, Anthropic product manager Cat Wu credits much of Claude Code’s popularity to the company’s AI models, which have gained favor among developers in recent years. Wu also notes that the Claude Code team intentionally adds elements of enjoyment to the product wherever possible.
Wu mentioned that Anthropic plans to continue expanding Claude Code’s availability, though the terminal will likely remain the primary platform for their AI coding solution.
“Looking ahead, one of our main priorities is ensuring that the CLI version remains the smartest and most adaptable way to use coding agents,” Wu said. “Still, we’re working to make Claude Code accessible everywhere, so it can support developers wherever they work. Expanding to web and mobile is a significant step in that direction.”
Anthropic states that its AI models are responsible for writing 90% of the Claude Code product itself. Wu, who previously worked as an engineer, says she now rarely writes code by hand, instead spending most of her time reviewing the outputs generated by Claude Code.
Earlier AI coding tools functioned mainly as autocomplete features, finishing lines of code as developers typed. The new generation of agent-based AI coding tools—like Claude Code—enables developers to create agents that operate independently. This evolution has led many software engineers to take on more of a supervisory role, managing AI coding assistants in their daily work.
Not all developers have embraced this change. A recent study revealed that some engineers actually experienced slower workflows when using AI coding tools such as Cursor. Researchers suggested this might be because participants spent significant time prompting and waiting for AI-generated code, rather than tackling other tasks. Additionally, AI coding tools can struggle with large or complex codebases, leading engineers to spend extra time resolving incorrect outputs from the AI.
Despite these challenges, companies like Anthropic are pressing forward with the development of AI coding agents. Anthropic CEO Dario Amodei predicted several months ago that AI would soon be responsible for writing 90% of code for software engineers. While this may already be the case within Anthropic, it could take longer for this trend to become widespread across the industry.
