Automated Code Remediation at Scale, 2nd Edition

Chapter 6. Beyond the Prompt: How Codebases Evolve at Scale

We are in a new era of software engineering—one defined by automation and AI.

Enterprise software isn’t a greenfield environment; it’s inherited, interconnected, and constantly evolving. Most teams manage large volumes of older code and technical debt spread across hundreds or thousands of repositories. These systems must adapt to a software supply chain in constant motion: vulnerabilities appear, libraries update, and frameworks deprecate APIs. Meanwhile, AI-assisted development is accelerating the creation—and duplication—of new code, adding even more surface area to maintain. In this environment, any meaningful code change must be applied consistently, at scale, and in compliance with an organization’s security, architectural, and quality standards.

The challenge is no longer the efficiency of writing code—it’s keeping up with it.

The Limits of AI Developer Assistants

Without a doubt, AI will play a significant role in the industrialization of software. For example, tools like GitHub Copilot have become powerful assistants for developers—helping them write net-new code faster, experiment with unfamiliar APIs, and speed up task completion in the IDE. But developer assistants alone aren’t built for the kind of large-scale, coordinated work enterprises must do to stay modern and secure.

They offer suggestive, nondeterministic outputs that are limited to a single file or repository. Their understanding of broader architectural ...