Is Claude Code Making Programmers Lonelier? Anthropic Engineer Finds a Fix

June 22, 2026

Has the influx of AI agents into coding made software development even more solitary than before? According to Fiona Fung, an engineering leader at Anthropic who works on the teams behind Claude Code and Claude Cowork, the answer may be yes.

In a recent interview on Lenny’s Podcast, Fung said her team noticed that as engineers increasingly relied on AI agents, they became more isolated. “After a while, we felt that coding became a lonely experience because people were working with their agents much more,” she said.

To address the issue, Fung sought to restore some of the collaboration that agentic coding had begun to erode. She started organizing programming lunches, hackathons, and blocks of shared “maker time” so engineers could work alongside one another and learn from each other’s AI workflows.

Fung added that while everyone on the team uses Claude Cowork, no two people use it in exactly the same way. Pair programming sessions, she said, allow team members to continuously learn new techniques and approaches from one another.

Claude has quickly become one of the most widely used AI products in software development. In a recent survey cited by Business Insider, Claude Code emerged as the dominant AI coding tool among startups, with some founders saying it has become their default choice for tackling complex engineering projects.

The shift reflects a broader transformation in software development, where engineers are spending more time directing AI agents, reviewing outputs, and managing parallel workflows. While AI-powered coding tools and “vibecoding” have made it easier for even non-technical founders to build software, many developers still see collaboration as essential for learning, creativity, and problem-solving.

Fung’s comments come as AI coding assistants become increasingly common across the software industry. Developers are spending more time managing AI agents, reviewing generated code and coordinating multiple automated tasks, changing how engineering teams work on projects.

To counter the isolation that emerged from agent-assisted coding, Anthropic began organising programming lunches, hackathons, and dedicated blocks of shared “maker time.” The initiative was designed to bring engineers together in person and allow them to observe how colleagues use AI tools in their daily workflows. According to Fung, developers often use AI systems differently, creating opportunities for teams to learn from one another when working side by side.