Rethinking Software Engineering Dimensions for the Agent Era

The author argues that a shift in software engineering research focus is changing the dimensions through which the field is analyzed. This shift demands a fundamental redesign of the roles and responsibilities among three key actors: humans, AI agents, and software systems themselves.

Consequently, when evaluating software engineering application scenarios, one must adopt a dual perspective: thinking for developers and thinking for agents . The author specifically calls out two analytical lenses that need re‑examination under this new paradigm:

Spatial structure analysis – the static architectural view of the system.

Evolutionary time analysis – the dynamic view of how the system changes over time.

For each lens, the article asks: what are the concrete application scenarios, and what value do they deliver?

The post includes three generations of diagrams that illustrate this thinking process:

A manually created summary diagram (author’s own synthesis).

A diagram expanded from the first using GPT (divergent thinking).

A diagram expanded using Gemini (alternative AI‑assisted expansion).

The images below are the original diagrams referenced in the article.

No quantitative benchmarks, code examples, or external citations are provided in the source; the contribution is a conceptual framework and a demonstration of using LLMs to broaden the initial analysis.