Cheap Code, Costly Judgment: A Case Study on Governable Agentic Software Engineering

Source

• Source name: arXiv
• Source kind: paper
• Authors / organization: James C. Davis, Paschal C. Amusuo, Tanmay Singla, Berk Çakar, Kirsten A. Davis
• Venue: arXiv: cs.SE
• Published: 2026-07-01
• Collected: 2026-07-02
• URL: http://arxiv.org/abs/2607.01087v1
• PDF: https://arxiv.org/pdf/2607.01087v1
• DOI: UNKNOWN

Plain-Language Brief

The authors of this paper study the shift in software engineering towards using Generative AI for low-cost code production and the challenges that come with it, specifically focusing on governance and maintenance. They present a case study of a 12-week development effort where an expert software engineer used AI agents to build a document accessibility remediation system.

What the Document Says

• The paper discusses the shift in software engineering towards using Generative AI for low-cost code production.
• The authors focus on the challenges of governance and maintenance in this context.
• They present a case study of a 12-week development effort using AI agents to build a document accessibility remediation system.
• The empirical record includes field notes, production code, tests, documentation, and agent tooling.
• From the record, they develop a middle-range theory of governance conversion, explaining how high-velocity agentic implementation becomes governable.
• The model explains how agentic implementation velocity surfaces recurring structural failure classes and how engineering judgment sustains velocity by converting those failures into durable governance mechanisms.
• The authors use their model to make testable predictions and describe implications for software engineering research and practice.

Key Concepts Mentioned

• Generative AI
• Software Engineering
• Governance
• Maintenance
• Agentic Implementation

Claims Made by the Document

• The shift in software engineering towards using Generative AI changes the central engineering problem.
• High-velocity agentic implementation surfaces recurring structural failure classes.
• Engineering judgment sustains velocity by converting failures into durable governance mechanisms.

Evidence Used

• 88 contemporaneous field notes, 420 KLOC of production code, and 1.16 MLOC of tests, lints, supporting documentation, and agent tooling

Methods / Approach

UNKNOWN

Limitations Stated by the Authors

UNKNOWN

Limitations Noted by the Agent

Agent interpretation — not from the source.

UNKNOWN

Open Questions

UNKNOWN

Notes for Future Q&A

May include interpretation. Labeled where it does.

UNKNOWN

Raw Abstract / Excerpt

Generative AI is shifting software engineering from a practice organized around scarce implementation effort toward one organized around abundant, low-cost code production. This shift changes the central engineering problem: not whether AI can generate useful code, but how engineers organize architectures, tools, evidence, and feedback loops so that AI-mediated development remains inspectable, correctable, and maintainable. We study this problem through a first-person case study: a 12-week development effort in which a single expert software engineer used frontier AI coding agents to build a document accessibility remediation system. The empirical record comprises 88 contemporaneous field notes, 420 KLOC of production code, and 1.16 MLOC of tests, lints, supporting documentation, and agent tooling. From this record, we develop a candidate middle-range theory of governance conversion, expressed as a process model explaining how high-velocity agentic implementation becomes governable. The model explains how agentic implementation velocity surfaces recurring structural failure classes, and how engineering judgment sustains velocity by converting those failures into durable governance mechanisms. In contrast to existing governance models that derive controls from known obligations, governance conversion explains how controls are discovered from failures that become visible only during agentic work. We use our model to make testable predictions and to describe implications for software engineering research and practice.