Wired spoke with Boris Cherny, head of Claude Code, to understand the secret behind the massive buzz surrounding the coding tool developed by Anthropic. The tool has not only achieved commercial success, but has radically changed how employees work within the company itself, heralding the dawn of the “agentic coding” era.
Article contents:
- Introduction
- Turning point: Opus 4.5 model
- Financial growth and commercial success
- Internal experimentation culture
- The evolution of the engineer’s role
- The future of smart agents
- Frequently asked questions
Introduction
In Silicon Valley, everyone is talking about Claude Code. What started as a simple auto-complete tool has transformed into a full “agentic” system, allowing developers to describe a feature in plain language and leave the rest to artificial intelligence. Boris Cherny, the man behind the product, asserts that they built the tool not for what AI was capable of back then, but for what it would be able to do in the future. And their bet appears to have paid off.
Turning point: Opus 4.5 model
Developers agree that the watershed moment was the release of the Claude Opus 4.5 model. Kian Katanforoosh, a lecturer at Stanford University, says this model represents a “qualitative leap” in coding capabilities, as it does not just feel like it is writing human code, but rather as if it found a better way to solve problems. This evolution has made Claude Code outperform strong competitors such as Cursor and Windsurf.
Financial growth and commercial success
The numbers confirm the technical success. In November, Anthropic announced that Claude Code reached $1 billion in annual recurring revenue in less than a year. By the end of 2025, this figure grew by another $100 million. Although the enterprise sector remains the largest, coding has become one of the company’s fastest-growing sectors, supporting its plans to reach positive cash flow by 2028.
Internal experimentation culture
What is interesting is the massive internal reliance on the tool. Cherny reveals that 100% of the code he writes personally, and 95% of the development team’s code, is generated by Claude Code. Even half of the sales team uses the tool weekly! This culture, technically known as “dog-fooding,” helped the company improve the product at breakneck speed and make it suitable for real and intensive use.
The evolution of the engineer’s role
Cherny compares this shift to the evolution from punch cards to modern programming languages. Engineers are not losing their jobs; rather, they are moving up the ladder of abstraction. Today, an engineer can run 5 or 10 software “agents” simultaneously via their phone or computer, each working on a different task, while they oversee and guide them. This eliminates tedious and routine work, leaving space for creativity.
The future of smart agents
The company’s ambitions do not stop at coding. It recently launched Cowork, which is Claude Code for non-programmers. This tool can manage files, send Slack messages, and organize schedules. Cherny believes the future will witness agents performing all routine life tasks, from filling out forms to transferring data, freeing humans to do what they truly enjoy.
Frequently asked questions
Question: What distinguishes the Opus 4.5 model?
Answer: It is considered the driving force behind the success of Claude Code, offering reasoning and programming capabilities that vastly outperform previous models.
Question: Do Anthropic employees use their own tool?
Answer: Yes, very intensively. Almost all code inside the company is now written with the help of Claude Code.
Question: What is the Cowork tool?
Answer: It is a version of the intelligent agent aimed at non-technical employees to automate office and administrative tasks.
