Having worked in the software engineering industry for three decades, I have had the privilege of experiencing several major paradigm shifts firsthand.
In the 1990s, software development was primarily about writing code. Requirements analysis was often informal, fragmented, and inconsistently specified. Consensus on system behavior was rare, and visual modeling was the exception rather than the rule.
Then came the Rational Unified Process (RUP). The pendulum swung decisively toward the early phases: requirements, analysis, and architecture. Significant effort went into understanding the problem space before a single line of code was written. Interesting enough, comparatively little was said about the hands‑on development activities themselves.
With the rise of Agile, the focus shifted again. Development moved back to center stage. Working software became the primary measure of progress, while requirements management and modeling were often treated as lightweight support activities, sometimes even as necessary evils.
Now, we find ourselves in a new era.
With AI and low‑code / no‑code platforms, something remarkable is happening: requirements are no longer just an input to development, they are becoming the means of development. Through prompting, a requirements analyst can now effectively “describe” behavior and let the system generate the implementation.
The pendulum is swinging again.
From development → to requirements → back to development → and now toward a convergence where expressing intent, behavior, and constraints clearly may be the most critical engineering skill of all.
It has been a fascinating journey to observe, and to be part of, this continuous oscillation between Requirements Management and Development.
And I suspect the next swing is already underway.
