An evidence-led five-part research series arguing that software engineering is being redefined as agentic engineering. The primary unit of work is shifting from hand-authored code to specifications, contracts, orchestrated agent activity, and verification of machine-produced artifacts.
Engineering judgment becomes the control layer: practitioners define objectives and constraints, compose agents and tools, govern effects, and decide what evidence is sufficient for acceptance.
| Part | Paper | Pages | |
|---|---|---|---|
| I | The End of Software Engineering | 26 | |
| II | The Anatomy of Agentic Engineering | 29 | |
| III | Plan, Act, Verify, Repeat | 24 | |
| IV | Agent Teams | 30 | |
| V | Context, Memory, and Governed Data | 28 |
This is a sourced practitioner series with a four-repository case study. Empirical sources support measured claims, and pinned code supports implementation claims. Productivity variation, quality risk, and review cost define the governance requirements created by the transition.
papers/latex/ LaTeX manuscript sources
papers/pdf/ publication PDFs
docs/papers/ HTML generated from the LaTeX sources
images/ title-page renders used as covers
reviews/ prose, technical, and formatting review records
team/ claim ledgers, citation checks, and integration records
website/ Next.js static publication site
scripts/ LaTeX-to-HTML conversion helpers
cd papers/latex
latexmk -pdf -interaction=nonstopmode -halt-on-error end-of-software-engineering.tex
cd ../../website
npm ci
npm run buildMatthew Long, The YonedaAI Collaboration, YonedaAI Research Collective Chicago, IL · matthew@yonedaai.com · https://yonedaai.com