New Issue: Orbital Catastrophe Ahead? Read Now

Software is becoming something you speak into existence

Coding for the rest of us finally feels possible now that tools like Claude Code turn plain English into working software

A hand holding a smartphone that displays the Claude app logo on the screen, with the Claude brand name blurred in the background.

A person holds a smartphone displaying the logo of “Claude,” an AI language model developed by Anthropic, with the company’s logo visible in the background.

Cheng Xin/Getty Images

Coding is having its GarageBand moment, its Excel moment. A moment like in 2004, when Apple made music production widely available. Or like in 1985, when Microsoft banished the idea of a spreadsheet as paper that required an accountant to calculate every value by hand.

Enter Claude Code, software created by the AI company Anthropic for “vibe coding”—writing code with natural language. Like GarageBand or Excel, Claude isn’t the first platform to try this, but it is the first to nail the anybody-can-do-this feeling.

Vibe coding platforms abound, and over the past years, I tried several, such as Windsurf and Replit. I was impressed by what they could do but often got stuck on problems that better coders could easily solve. And building websites that matched what I hoped for was challenging.


On supporting science journalism

If you're enjoying this article, consider supporting our award-winning journalism by subscribing. By purchasing a subscription you are helping to ensure the future of impactful stories about the discoveries and ideas shaping our world today.


What makes Claude Code different is the ease of use and the speed at which it understands problems. I’ve created several websites with the latest version, and each time I hit a snag, I described the problem, and Claude Code fixed it. I’ve used it through different interfaces, and the app offers the most user-friendly experience for “muggles”—folks with no coding magic.

Naturally there is a rival. One can’t talk about Claude Code without mentioning OpenAI’s Codex. OpenAI was the first company to popularize a large language model (LLM) that turned plain-English instructions into code. GitHub’s Copilot launched in June 2021, but it was built on Codex, which OpenAI released that August.

In 2023 came the plot twist: Codex, which had been steadily evolving, was discontinued. An OpenAI e-mail notice encouraged developers to use GPT-3.5 instead. Coding, it seemed, would be just one skill among many that the AI system would perform.

That’s when Anthropic entered the race with a different vibe. In March 2023 it introduced Claude as a model trained to be “helpful, honest and harmless.” Claude could code, but the early praise wasn’t about fireworks; it was about feel. Users liked that exchanges “feel like natural conversation,” Autumn Besselman of Quora said in a statement from Anthropic.

Only two months later, Anthropic announced it had “expanded Claude’s context window from 9K to 100K tokens,” which corresponded to about “75,000 words.” A context window is basically the desk space that allows you to spread out your code and documentation. When Anthropic fed Claude the entire novel The Great Gatsby with one line changed, the model spotted the alteration in 22 seconds.

Over the next two years, Claude’s coding skills dramatically improved. By early 2025, using AI to code was gaining popularity among noncoders, and that February Andrej Karpathy, a former OpenAI researcher, coined the term “vibe coding.”

In May 2025 OpenAI relaunched Codex as “a cloud-based software engineering agent that can work on many tasks in parallel” and made it available in ChatGPT. It also made Codex CLI open source, “fostering more developer goodwill,” according to TechCrunch, thanAnthropic, which issued a takedown notice to a developer who had reverse-engineered Claude Code.

Tensions peaked last August, when Wired reported that Anthropic claimed OpenAI staff were using Claude Code ahead of the launch of GPT-5. Anthropic revoked OpenAI’s access in accordance with its commercial terms, which bar customers from using Claude to “build a competing product.” An Anthropic spokesperson told Wired, “OpenAI’s own technical staff were also using our coding tools.”

By January 2026, on X, Karpathy described a “phase shift in software engineering,” admitting, “I really am mostly programming in English now,” while Boris Cherny, an Anthropic staffer, wrote, “Pretty much 100% of our code is written by Claude Code + Opus 4.5.” According to Fortune, an Anthropic spokesperson said that Claude Code now writes about 90 percent of its own code.

Though both Codex and Claude Code are powerful, in my experience, Claude Code simply builds websites faster, even fairly complex ones requiring interactive 3D presentations.

Memes have sprung up, evoking the pleasure of using it. People have cited “Claudeholism: The New Addiction of the Century” or asserted that “Claude Code is the opium of the permanent underclass” in reference to the software’s ability to remove the pain of drudgery. And then there’s the cuteness of its progression indicator—verbs like crunching, moseying, perusing, reticulating, discombobulating and zesting. When Claude Code works, it’s sparkling or simmering. Silly as this is, it’s hard not to love.

Deni Ellis Béchard was formerly Scientific American’s senior writer for technology. He is author of 10 books and has received a Commonwealth Writers’ Prize, a Midwest Book Award and a Nautilus Book Award for investigative journalism. He holds two master’s degrees in literature, as well as a master’s degree in biology from Harvard University. His most recent novel, We Are Dreams in the Eternal Machine, explores the ways that artificial intelligence could transform humanity. You can follow him on X, Instagram and Bluesky @denibechard

More by Deni Ellis Béchard

Subscribe to Support Independent Journalism

Great science journalism requires human expertise, time, effort and creativity. And it costs money. That’s why I and the journalists here at Scientific American hope you’ll join our community.

When you subscribe, you are supporting staff and freelance journalists who are passionate about telling science stories that are true, important and compelling. Our editors and reporters are often experts in their fields, which means they understand the nuances of big discoveries and can untangle the breakthroughs from the hype. With a subscription, you are also supporting rigorous fact-checking to ensure the words we publish are precise and accurate. And you’re supporting original illustrations, graphics and photos that bring you closer to an advanced laboratory, an ice sheet in Antarctica or a space mission in orbit. You’re helping us craft other types of high-quality journalism as well: Our newsletters are carefully written, edited and curated by staffers you have or will come to know and love. Our Science Quickly podcast is based on original reporting, collaboration with editors and scientists and exacting production.

Subscriptions keep this engine running so we can continue to deliver thoughtful, rigorous and independent science journalism to you. In an era of viral misinformation, this work is crucial. If you value what we do, I hope you’ll consider joining us as a subscriber

Thank you,

Jeanna Bryner, Editor in Chief, Scientific American

Subscribe