Software engineers may be paying the price for Silicon Valley’s token maxing frenzy.
As companies encourage developers to use AI coding tools in hopes of improving productivity, a rift is emerging within some engineering teams, said Deedy Das, a partner at Menlo Ventures, a venture capital firm that invests in AI and enterprise software companies.
The first are what he calls “lazy” engineers. These are workers who rely heavily on AI to write code, answer questions, prepare updates, and complete tasks with minimal effort.
And then there are the “artisans,” experienced engineers who shoulder the burden of understanding, reviewing, and fixing the ever-growing flood of AI-generated code.
“Most software engineers are facing an identity crisis bordering on depression,” Das said in an X post over the weekend.
“The craft they loved is dead.”
Das’ comments tap into the growing debate about how AI coding tools are reshaping software engineering, a central theme of Business Insider’s latest series, “The Great Coding Reset.”
As companies encourage developers to use AI to generate more code, some engineers say their roles are shifting from writing software to reviewing, managing and validating machine-generated work, a shift that is forcing them to rethink where their value lies.
The broader issue is what Business Insider’s Amanda Huber recently described as “AI sprawl.” This means employees are juggling multiple AI tools, duplicating work, and producing ever-increasing output without clear evidence that the company is significantly more efficient.
“The artisans are tired. Very tired,” Das wrote. “The burden of review is entirely on the craftsman. It’s the burden of understanding.”
As AI makes code generation easier, reviewing and maintaining that code has become a new bottleneck.
This change reflects the rise of “botsitting”, which monitors AI, corrects mistakes, and verifies its work.
Das previously said that many software engineers feel like their “lifelong skills are becoming useless” amid the AI boom, but said some engineers now find themselves with more and more bugs creeping into production environments, often buried in pull requests.
“The ship they loved is dead,” he wrote.
Das said that while many companies have successfully integrated AI into software development, this tension is common in large organizations where AI-generated artifacts are growing faster than teams can evaluate them.
“This tends to happen in large companies that have been around for more than 10 years, where there is a lot of variation in talent,” he said. “But it happens a lot.”