As the tech industry continues to scale AI use, some companies are hitting snags not in the technology itself, but in the human workforce developing and working alongside AI agents.
“The other thing that we found interesting on the Claude Code team is, after a while, we felt it could start being a lonely experience because we all started just working with our agents so much,” she said.
Anthropic began implementing hackathons “just to make sure we’re interacting together as a team,” as well as pair programming lunches for employees to share how they’re using Claude Code, Fung said. She deemed both interventions successful.
“When we do pairwise programming, we actually learn so much from each other,” she said. “Every time I watch someone work, I learn something myself as well.”
An Anthropic spokesperson said the company is paying close attention to how its AI tools impact how employees work together.
“We’re seeing engineers find new ways to learn from and build alongside one another, in what’s really an evolution of pair programming,” the spokesperson said in a statement to Fortune. “Where pairing was once about working through a tough problem together, it’s now increasingly about seeing how a colleague uses these new tools and systems differently than you might, so even as more of the work shifts toward collaborating with agents, engineers keep learning from one another. Sharing how our own work is changing, including the hard parts, helps us build tools that best serve the people using them.”
“We’ve undermined the trust you have that your specific expertise and contribution will be valued, that you will grow and advance your career, and that this will be a place where you can actually have an impact,” Bosworth wrote. “We shook up the management structure that was providing you stability while rapid changes in strategy, including the boom/bust cycle of hiring, left entire teams in the lurch.”
Bosworth said the company will take steps to be “fun and enjoyable” for employees, such as increasing travel budgets and social event spending, as well as “improving microkitchens” where workers can take breaks and eat snacks.
“I hope we can rekindle the best of the culture we joined,” Bosworth wrote.
Jeffrey Pfeffer, a professor of organizational behavior at the Stanford Graduate School of Business, said employee unrest in the tech industry more broadly is a result of the “move fast and break things” tradition that often leaves workers vulnerable.
“The tech industry has said people are the most important asset, but they never act that way,” Pfeffer told Fortune. “Many people, I think, still go into jobs believing that their employers care about them and will take care of them, and so they are of course disappointed when they find that their employers actually don’t care about them or will take care of them.”
“We know that as tools come in, it often restructures jobs: Certain tasks disappear, other tasks appear. That is a stressful time, right?” Neil Thompson, an assistant professor of innovation and strategy at the MIT Sloan School of Management, told Fortune. “People don’t know what those new tasks are going to look like. They can see old tasks being threatened, and that can also make it more intimidating, particularly when you have quite impressive results coming out.”
“On days where everything works well, I can’t help but think nothing I do matters, everything is automated and better and faster than I ever will be,” the employee said. “But then there are days where everything breaks and I don’t understand why and I realize I have no idea what I’ve been up to anymore.”
The report also published comments from employees who said Claude has augmented their work and left humans in the driver’s seat of selecting the direction for models to move. Anthropic touted Claude’s ability to ship more than 800 application programming interface (API) error fixes in April, a task the company said would have taken a human four years to complete.
There are ways to ease the feeling of threat some workers have in navigating AI advancements, Thompson said. Companies can prepare workers through reskilling, but also by being transparent in changes that may happen to labor as a result of automation, both for better and for worse. Historically, Thompson said, this has looked like fewer experts in certain positions, with wages rising for these workers, and in other areas, the number of roles expanding, with wages decreasing.
“That’s a tough thing, but often there are more opportunities for people to enter that field, because now more people are capable of doing that,” Thompson said. “If we think about the morale of people in these areas, I think one thing that can help is to say, like, this is not all doom and gloom.”
