We are living through a moment where advances in AI have made the delivery and implementation of software systems and features significantly faster. I have seen this in my own work. But when the most laborious parts of writing code can be dramatically reduced, the goalposts naturally shift not only for clients but for software developers too.
Building software has always been an interactive and evolving process. From one version to the next, both the software itself and the expectations around it continue to change. With AI, however, even the baseline expectation has moved. As a developer, you may have developed a knack for estimating how long a particular feature or piece of work should take. That intuition is often built over years of execution. AI has disrupted that internal compass. In some cases, it can leave you feeling as though you have not done enough, simply because the current pace no longer aligns with the expectations you formed through experience. And that misalignment can create a subtle pressure to keep doing more, even when the amount already accomplished is substantial.
What would once have been considered version 1 of a system, enough to qualify as an MVP, now feels closer to an initial functional demo or prototype. In many cases, what used to resemble version 5 now appears to be the new version 1. That shift is understandable, especially given how the AI revolution has been marketed. And as these technologies continue to advance, the pace at which the goalpost moves will likely accelerate beyond what many professionals can sustainably keep up with.
At the same time, software is increasingly being designed with agentic flows in mind. That means systems must, at their core, ensure that all essential functions can still be carried out by a human as a redundancy, even if agentic flows serve as the primary operational mechanism. Fallback mechanisms and activity tracing are no longer optional; they are becoming defining characteristics of modern software. In this model, the human in the loop becomes the “pilot” of what are meant to be semi-autonomous systems.
Having a strong grasp of the fundamentals, and a well-developed instinct for systems thinking, has never mattered more. As execution becomes faster and more automated, context becomes the real differentiator. Clear thinking, sound judgement, and the ability to articulate intent precisely are what will determine how effectively software is built. The edge is no longer just in producing faster, but in knowing what to build, why it matters, and how it holds together as part of a larger system.