Is AI coding actually making our professional identities
We used to talk about "Prompt Engineering" or the need for complex Model Context Protocol (MCP) setups to get decent results. I spent a significant amount of time building custom tools just to bridge the gap between my intent and the model's output. But that era is dying. Whether you’re using Claude or the latest GPT iterations, the need for granular, step-by-step instructions is vanishing. Even "graph engineering" techniques that people are currently hyping up seem to be getting baked into the models natively. The moment I master a new AI engineering technique, the next model update renders it an out-of-the-box feature. It feels like running on a treadmill that keeps accelerating.
The real blow to my ego, however, isn't the speed—it's the structural quality.
The AI is consistently better at implementing sophisticated architectural patterns than I am. It applies ADT patterns, immutability, Result/Option patterns, and monadic chaining with a level of meticulousness that I struggle to match. When it comes to structuring something using Hexagonal architecture, MVVM, or MVI, the AI doesn't "forget" a layer or take a shortcut. It just executes.
I spent years memorizing GoF patterns, studying how to tweak them into a personal style, and learning how to balance abstraction with readability. Now, I watch a model generate a perfectly structured, highly testable codebase in seconds. For the vast majority of freelance work—which is largely business logic and CRUD operations—the AI is simply more capable. It handles domain modeling and boundary testing with a precision that makes my "deep thinking" feel redundant.
I used to need a second monitor for documentation and a dozen Chrome tabs for Google searches just to implement a clean pattern. The AI does that instantly.
This leads to a very grim realization for those of us who aren't working at the bleeding edge of Big Tech research: if the AI can handle the architecture, the domain modeling, and the implementation of complex patterns better and faster than I can, what exactly is my role? I'm no longer a "builder" in the traditional sense; I'm an assembler of business logic on top of frameworks that the AI understands better than I do.
The "vibe coding" aspect is fun, don't get me wrong. But there is a massive difference between enjoying a new way to work and feeling the actual floor fall out from under your competitive edge. How are the rest of you handling this? Are we just transitioning into high-level supervisors, or are we actually losing the very skills that made us developers in the first place?