During a recent experience, Pietro Schirano, the founder of an AI product design startup MagicPath and former Anthropic engineer, engaged with a unique AI system to analyze his raw DNA data from Ancestry DNA. He employed Claude Code—an AI agentic coding tool developed by Anthropic—to carry out the complex task. Instead of a single analysis, Claude Code replicated itself across his computer, spawning distinct sub-agents specialized in different genomic areas, including cardiovascular health, aging processes, and autoimmune conditions.
The AI’s revelations aligned strikingly with Schirano's personal experiences, such as his anecdotal ability to metabolize caffeine effectively compared to peers, a trait humorously attributed to his Italian heritage. Claude Code confirmed this genetic disposition and uncovered a genetic predisposition to Alzheimer's disease, subsequently recommending supplements tailored to his DNA profile.
Claude Code was introduced in February 2025 as Anthropic’s pioneering AI agent—designed to go beyond simple user interaction. Instead of merely chatting, it actively accesses files and runs programs directly on the user’s computer. Its sub-agents can dissect and process specialized data sets autonomously. Despite these capabilities, Claude Code has largely remained a tool for enthusiasts comfortable with a traditional command line interface—a mode of user interaction that has mostly faded from the mainstream use over recent decades.
This accessibility barrier is now set to be addressed with the launch of Claude Cowork. Announced on a recent Monday, Claude Cowork essentially functions as ‘Claude Code for the rest of your work,’ aiming to transform the AI’s power so that non-coders can harness it efficiently. Martin DeVido, a developer who used Claude Code in an experiment growing a tomato plant autonomously, expressed enthusiasm for how this new interface could astonish many users unfamiliar with coding.
Claude Cowork offers a more user-friendly interface, obscuring some of the technical challenges that previously limited Claude Code’s broader adoption. Currently, the tool is accessible as a research preview for customers subscribed to Anthropic’s Max plan at $200 per month. The release, although promising, is candidly described by Felix Reiseberg, lead engineer on the project, as rough around the edges.
Users have reported difficulties such as unsettling error messages and connectivity issues with their calendars, signaling the nascent stage of the product. Notably, Claude Cowork was developed in less than two weeks, mainly through contributions from Claude Code itself. This rapid, AI-assisted software development cycle represents a new paradigm where AI tools are not only end-products but also active participants in their own creation—a process sometimes accompanied by instability.
From its inception as a demo, Claude Code quickly progressed to generate annualized revenue reaching one billion dollars within six months, forming a significant portion of Anthropic’s income at the time. The success of these agentic AI tools illustrates a broader momentum toward empowering AI systems to perform autonomous actions rather than merely responding conversationally.
Claude Code and Claude Cowork are part of a growing group of AI technologies designed to transition chatbots into actionable agents. Other notable agentic coding tools like Cursor and OpenAI’s Codex have gained traction among coding professionals, while AI-powered browsers such as ChatGPT Atlas and Perplexity’s Comet provide autonomous web interaction capabilities. According to Jean-Stanislas Denain, an AI researcher at Epoch AI, tools like Claude Code underscore this ongoing shift in AI functionality.
However, questions remain about Claude Cowork’s versatility outside of coding tasks. The model powering it, Opus 4.5, excels in programming but has demonstrated limited performance on a broader range of knowledge work assignments. Research from the Center for AI Safety’s Remote Labor Index revealed that Opus 4.5 successfully completed only nine out of 240 diverse freelance projects, which included complex tasks such as architectural planning and video game development. The study highlights that specific cognitive limitations remain unaddressed by Claude Code.
Additionally, the initial deployment of Claude Cowork to a limited number of customers makes it difficult to assess how efficiently individuals lacking coding expertise can benefit from the tool. Despite this, the move by Anthropic reflects the anticipated evolution from chat-based AI interactions toward AI agents capable of executing tasks independently—a transformation foreseen since well before ChatGPT’s emergence.
Daniel Kokotajlo, executive director of the AI Futures Project, positions Claude Code as part of a broader long-term trend that is expected to continue developing over the coming years. Kokotajlo previously predicted an era termed the 'age of the AI assistant' would begin around 2026, with subsequent predictions pointing to the arrival of artificial superintelligence by 2034.
For users like Schirano, these AI advancements promise significant economic impacts, especially within white-collar professions. He anticipates Claude Cowork may cause a larger disruption in economic indices than any previous technology due to its potential to reshape professional workflows.