Not their job
A QA engineer doing development work. A product owner fixing bugs in code. Nobody told them to. Here is why it happened.
Shekhar was afraid to switch from .NET to Java.
Not afraid in the abstract. Afraid in the practical sense: he had built his identity around one stack, and moving to another meant starting over. Slower. Asking questions he should already know the answers to. Possibly making the team wait while he caught up.
That is the real constraint behind “I’m a .NET developer, not a Java developer.” It was never about capability. It was about the cost of starting somewhere new.
Four weeks later, he was troubleshooting a React Native mobile application on his own. Not Java. React Native.
I heard about this from Archana and Teesha, the coaches who ran our AI in SDLC workshop with his team at a software services company. I had worked with Shekhar in the first few days, and I could picture exactly where he had started from.
The question that stayed with me: what changed?
Here is how it started. About two weeks into the workshop, a defect came up in one of the smaller legacy systems his team was responsible for. He had been working with AI for two weeks. He thought: why not give it a try. He asked Claude to investigate the issue, come up with a plan to fix it, and explain how it would test the result. He reviewed it at a high level, then asked it to implement. It worked. He came back the next morning, got the fix reviewed by another developer on the team, and that was it. A new technology. One evening.
That experience changed something. He now says he wants to stop seeing himself as a .NET developer only.
Not the skill. Four weeks is not long enough to become a React Native developer. What changed was the cost of starting.
When you are learning something unfamiliar without AI, you slow the team down. You ask questions that take a colleague out of their flow. You make mistakes that have to be unpicked by someone who knows the system better. The cost is visible, and it falls on the people around you. So you stay in your lane. You learn the language of that lane and defend it: “I am a Java developer.”
With AI, the cost of starting collapses. You can ask what feels like a stupid question without it feeling stupid. You can explore an unfamiliar codebase at your own pace, without anyone watching. You can make a mistake and ask AI to help you understand what went wrong, without it affecting the next sprint.
Shekhar was not the only one. Sachin, a QA engineer, started working on development tasks he had never touched before. Naseem, a product owner, started fixing bugs in code herself. Developers began learning DevOps from Nishant, voluntarily, because they could pick up unfamiliar territory without it costing the people around them.
The silos did not dissolve because anyone decided to dissolve them. They dissolved because the economics changed.
This matters for leaders in a specific way. If your engineers are staying in their lanes, the instinct is to assume it is a culture problem: they are territorial, resistant, not collaborative. Sometimes that is true. But more often it is a cost problem. They are rational. Going outside your lane is expensive, for you and for the people around you. So you do not.
When AI makes that cost near-zero, the rational choice changes.
The identity labels were not about what they could do. They were about what they were willing to start.