At the very forefront of AI development, coders are now relying fully on AI to do their daily jobs.
Boris Cherny, the creator of Claude Code at Anthropic, recently disclosed that he didn’t open an integrated development environment (IDE) even once during the past month. Instead, every line of code he shipped—259 pull requests totaling 497 commits, 40,000 lines added, and 38,000 lines removed—was written entirely by Claude Code powered by Opus 4.5.
“The last month was my first month as an engineer that I didn’t open an IDE at all,” Cherny wrote on X. “Opus 4.5 wrote around 200 PRs, every single line. Software engineering is radically changing, and the hardest part even for early adopters and practitioners like us is to continue to re-adjust our expectations.”
The shift represents a dramatic acceleration in AI-assisted development. In May 2025, Cherny had noted that 80% of Claude Code’s codebase was already being written by Claude Code itself—a remarkable milestone suggesting the tool had become proficient enough to build and maintain its own infrastructure. Now, just months later, that figure appears to have reached 100% for Cherny’s personal workflow.
The Mental Shift Required
Cherny emphasized that the biggest challenge isn’t the technology itself, but rather the mental adjustment required to trust AI capabilities that continue to expand rapidly. He described how ingrained habits from years of traditional coding can actually hinder effective use of modern AI tools.
“Sometimes I start approaching a problem manually, and have to remind myself ‘Claude can probably do this,'” he explained, recounting a recent debugging session for a memory leak in Claude Code.
While Cherny initially approached the problem using conventional methods—connecting a profiler, using the app, pausing the profiler, and manually examining heap allocations—a coworker took a different approach. That colleague simply asked Claude to create a heap dump and analyze it for retained objects. Claude identified the issue on the first attempt and submitted a pull request with the fix.
“The same thing happens most weeks,” Cherny noted, highlighting how frequently AI now outperforms traditional debugging methods.
New Developers Have an Advantage
Interestingly, Cherny observed that newer engineers and recent graduates may actually have an advantage in this AI-powered environment. Without legacy assumptions about what AI models can and cannot do—mental models formed during the era of less capable systems—they’re able to leverage the technology more effectively.
“It takes significant mental work to re-adjust to what the model can do every month or two, as models continue to become better and better at coding and engineering,” he said.
This observation suggests that the rapid pace of AI advancement is creating a unique situation where experience with older tools may actually create blind spots, while newcomers approach each problem with fresh eyes and full confidence in current AI capabilities.
Coding as Code Review
The transformation Cherny describes aligns with predictions he made earlier this year. In June 2025, he suggested that the future of coding would be less about writing code and more about reviewing it—a shift that appears to be materializing faster than many anticipated.
With Claude Code now capable of running autonomously for “minutes, hours, and days at a time” using stop hooks, the traditional image of a developer typing away at a keyboard is giving way to a new paradigm: engineers as orchestrators and reviewers of AI-generated code.
Still the Beginning
Despite these dramatic changes, Cherny maintains that the industry is only at the start of this transformation. “Software engineering is changing, and we are entering a new period in coding history,” he wrote. “And we’re still just getting started.”
His experience offers a glimpse into what may soon become standard practice across the software industry. If one of the creators of advanced AI coding tools has completely transitioned away from traditional IDEs in favor of AI-assisted development, it raises the question: how long before this becomes the norm for software engineers everywhere?
For now, Cherny’s month without an IDE stands as both a personal milestone and a signal of the profound changes reshaping one of the technology industry’s most fundamental activities: writing code.