Anthropic's Cross-Platform Play: What Actually Ships Today
While competitors focus on chatbots and APIs, Anthropic has quietly built a multi-surface product line: Claude Code across five platforms, mobile apps with 109K ratings, and computer use that lets Claude operate your Mac.
Anthropic has built something its competitors haven’t: a shipping product line that extends beyond the browser. While OpenAI and Google offer chatbots with API access, Anthropic now sells Claude Code across five surfaces, mobile apps on iOS and Android, a Chrome extension, and computer use capability that lets Claude operate a Mac the way a person would.
Claude Code Everywhere
Claude Code began as a terminal tool. It now runs in VS Code, JetBrains IDEs, a standalone desktop app, and a web version at claude.ai/code—five entry points for the same agentic coding assistant.
The desktop app, available for macOS (Intel and Apple Silicon) and Windows (x64 and ARM64), adds capabilities the terminal can’t match: visual diff review, multiple sessions running simultaneously, scheduled recurring tasks, and cloud sessions that continue working when your laptop sleeps. The web version requires no local installation, with long-running tasks that complete in the background.
“Claude Code is an agentic coding tool that reads your codebase, edits files, runs commands, and integrates with your development tools.”
The JetBrains plugin brings inline diff viewing and selection context sharing to IntelliJ IDEA, PyCharm, and WebStorm. Slack integration exists for teams that want Claude Code responses in their messaging workflow. All of it requires a paid subscription—Pro at $17/month (annual) or Max starting at $100/month.
Mobile With Substance
The Claude iOS app carries a 4.7-star rating from 109,000 reviews—numbers that suggest actual daily use, not novelty downloads. Features go beyond chat: voice dictation in multiple languages, visual analysis of photos and PDFs, web search with citations, and integrations with Google Drive, Gmail, and Calendar.
The app runs on iPhone, iPad, Mac (native), Apple Vision, and Apple Watch. Android availability is confirmed on the pricing page, though without a comparable primary source listing features.
Computer Use: The Experimental Edge
In October 2024, Anthropic shipped computer use as a public beta—the first frontier AI model to offer the capability. “Instead of making specific tools to help Claude complete individual tasks, we’re teaching it general computer skills,” the announcement read.
Eighteen months later, computer use remains in “research preview” status, restricted to macOS, and available only on Pro and Max plans—explicitly not on Team or Enterprise tiers. The capability lets Claude open applications, read screen content, click buttons, and type text. A practical example from the documentation: Claude can compile a Swift app, launch it, click through every control to verify it works, and screenshot the result, all in a single conversation.
Safety constraints are built in. Browsers and trading platforms are view-only. Terminals and IDEs are click-only. Other applications get full control with per-session approval. Screenshots are downscaled before transmission. Press Escape anywhere to abort.
On the OSWorld benchmark, Claude scored 14.9% in screenshot-only mode, nearly double the 7.8% of the next-best AI system. The number sounds low until you consider the benchmark measures multi-step computer tasks humans do daily.
The Browser Extension Question
Claude for Chrome appears in the pricing tier feature list as a subscription benefit, and “Chrome extension (beta)” shows in Claude Code documentation navigation. But dedicated documentation for the extension didn’t surface in this research. What it does beyond “Claude Code in Chrome” remains unclear.
What to Watch
Anthropic’s cross-platform push is real shipping product, not announcement vapor. The question is whether computer use graduates from research preview to production capability—and whether the explicit exclusion from Team and Enterprise plans signals technical caution or a deliberate consumer-first strategy. The company that first makes agentic computer control reliable for enterprises will have built something genuinely new.