Editorial illustration for Anthropic's Claude Code now writes most code; Claude Cowork set for enterprise
AI Writes 90% of Claude's Code: Anthropic's Big Leap
Anthropic's Claude Code now writes most code; Claude Cowork set for enterprise
The percentage is absurd. Companies using Claude Code now have it write ninety percent, sometimes all, of their software. Projects that stretched over fiscal quarters now ship in weeks.
This isn't a better autocomplete. It’s a new physics for how fast things get built.
But the coding is just the proof. The actual shift is architectural. Anthropic is tearing down the chatbot. They’re turning their model into plumbing.
Their Model Context Protocol, MCP, is the pipe. It lets Claude reach directly into a company’s live systems—Slack, Google Drive, Salesforce, financial databases—and pull information without a human acting as a go-between. It connects on its own.
This week, Anthropic announced Claude Cowork, an enterprise system built on that protocol. It’s not a bot you talk to. It’s a layer of reasoning installed across a company’s entire digital body.
The trajectory began with Claude Code, which Jensen noted had taken coding use cases "from assisting on tiny tasks to AI writing 90 or sometimes even 100% of the code, with enterprises shipping in weeks what once took many quarters." But the deeper structural shift was the adoption of MCP -- the Model Context Protocol -- which has become the connective tissue allowing Claude to reach into and act upon data across an organization's entire technology stack. Where previous AI tools were constrained to the information users manually fed them, MCP-connected Claude can pull context from Slack threads, Google Drive documents, CRM records, and financial systems simultaneously. This is what makes the plugin architecture announced Tuesday fundamentally different from earlier chatbot-style enterprise AI: it turns Claude into a reasoning layer that sits across an organization's existing infrastructure rather than alongside it. The implications for the broader AI industry are profound.
The goal is to dissolve the work itself. Right now, knowledge work is mostly shuttling context between apps. You read an email, paste a number into a spreadsheet, summarize a Slack thread for a report.
Claude Cowork, if it works, wants to erase that middle layer. The AI becomes the system you work through, not a tool you use.
For a company, the advantage stops being about doing old tasks faster. It becomes about making decisions at the speed your data can be connected. The model sees the Slack argument, the updated sales figures, and the project timeline all at once.
It can propose a solution. The competitive edge will belong to the organizations built to be read and acted on by this kind of intelligence. The alternative is being left behind by those that are.
Related Reading
Browse all Anthropic news →
Business & Startups Nordic pilot adds Gemini for Education, NotebookLM to boost AI literacy
Business & Startups Kling launches Video O1, all-in-one model with MVL bridge using transformer
Business & Startups DeepSeek Seeks More Capital Weeks After USD 7B Funding Round
Related Topic Trump cracks down on Anthropic after Amazon tip; staff largely foreign
Open Source
Related Topic Anthropic says Claude writes >90% of its code; AI pause button urged
LLMs & Generative AI
From Archives Nimble unveils Agentic Search Platform with 99% accuracy, 3.2M interactions
From Archives IBM's USD 40B loss tied to COBOL translation, not true modernization
From Archives The Vergecast on Claude's vibe coding, email safety, and phone upgrade timing
LLMs & Generative AI
From Archives Pentagon seeks to label Anthropic a national security risk, limiting partners
Open Source
Common Questions Answered
How much code is Claude now generating at Anthropic?
According to the article, Claude is now generating nearly 100% of the code at Anthropic, with engineers shipping pull requests containing 2,000 to 3,000 lines fully generated by Claude. [medium.com](https://medium.com/@manojreddy_83742/i-built-an-mcp-server-so-claude-writes-code-like-my-team-not-like-a-tutorial-4d7adfc21587) reports that some engineers, like Boris Cherny, have not manually touched code for over two months.
What is the Model Context Protocol (MCP) and how does it improve AI code generation?
The MCP is a system that allows Claude to access and understand an organization's entire technology stack, enabling more contextually accurate code generation. [medium.com](https://medium.com/@manojreddy_83742/i-built-an-mcp-server-so-claude-writes-code-like-my-team-not-like-a-tutorial-4d7adfc21587) describes it as a way to index repositories at the function level, detect design patterns, and provide metadata filtering to generate code that matches a team's specific coding standards.
What productivity improvements has Anthropic seen with AI-generated code?
Internal data shows a 50% increase in productivity and 2-3x faster work compared to traditional coding methods. [twinstrata.com](https://www.twinstrata.com/news/claude-writes-code-at-anthropic/) notes that Claude Code can now run 20 independent actions per task, which is double the number from six months ago, significantly accelerating software development processes.
Further Reading
- Anthropic’s Claude Code Security rollout is an industry wakeup call — CSO Online
- Anthropic's New AI Model Targets Coding, Enterprise Work — Campus Technology
- Claude Code Security and the Future of AI-Driven Cybersecurity — BISI
- Making frontier cybersecurity capabilities available to defenders — Anthropic
