Anthropic Engineering Lead: Claude Code Is Making Programmers More Lonely
Programming has always been a profession prone to isolation, but the rise of AI coding agents is deepening that solitude in unexpected ways. According to Fiona Fung, the Anthropic engineering lead responsible for the Claude Code and Claude Cowork teams, engineers who lean heavily on AI agents end up talking to each other far less than they used to.
Fung put it plainly: “Over time, we noticed this way of working can feel lonely, because everyone is spending so much time working with their own agent.” The observation comes from inside Anthropic itself, where Claude Code has become central to daily engineering workflows. As the tool gained traction, the team saw a measurable drop in spontaneous collaboration — fewer quick conversations, fewer whiteboard sessions, fewer moments of shared problem-solving.
To reverse the trend, Anthropic began deliberately engineering social contact back into the workday. The company organized programming lunches, hackathons, and dedicated co-working blocks where engineers sit together and learn from each other’s AI workflows. “Everyone uses Claude Cowork, but the ways they use it vary dramatically. When we do pair programming, we learn a huge amount from watching each other,” Fung noted.
Claude Code has rapidly become one of the most widely adopted tools in software development. A survey of over 20 startup founders and venture capitalists found that Claude Code is now the most-used AI programming tool among startups. Several founders reported reaching for Claude Code first when tackling complex engineering challenges — ahead of any alternative.
The shift in how engineers spend their time is structural. More hours are going into orchestrating agents, reviewing generated output, and juggling multiple parallel tasks. Meanwhile, “vibe coding” — developing software through natural language prompts alone — is exploding in popularity. The barrier to building custom tools has dropped so dramatically that non-technical founders can ship functional products without ever assembling a traditional engineering team. The “solo founder” is becoming a norm, not an exception.
Yet building everything alone comes with its own cost. Many founders still insist that collaboration is essential, and Fung agrees. Even when team members use AI in radically different ways, Anthropic is pushing to create more opportunities for engineers to work shoulder to shoulder. As Fung put it: “Every time I watch how someone else works, I learn something myself.”