CBD-Centered Development Process in the AI Era

The Unified Process presented three important principles: iterative and incremental development, architecture-centric design, and use-case–driven development.

These principles remain valid today and continue to serve as a useful basic framework for organizing projects in the AI era.

UP also assumed a component-based architecture, and in that sense it was a development process oriented toward Component-Based Development (CBD).

However, what UP mainly provides is a framework for project management and the development process.

While it assumes the concept of CBD, UP does not explicitly position it as the central structural model of the development process when it comes to how software systems should be constructed.

In the era when UP was originally proposed, software structures were typically designed using units such as the following.

Human programmers were generally responsible for implementing these structures.

However, once AI-based code generation becomes widespread, the situation changes significantly.

AI can generate code very quickly, but it is not particularly good at autonomously stabilizing the overall structure of a system. When structural constraints are weak, generated code can easily diverge, causing the overall system consistency to collapse.

Therefore, in the AI era of development, elements such as the following become more important than individual implementation techniques.

All of these relate to architectural constraints.

A framework that provides such structural constraints is Component-Based Development, which becomes increasingly important in the AI era.

In the AI era of software development, stabilizing the overall system structure becomes more important than the implementation itself.

At the center of this approach is Component-Based Development (CBD).

In CBD, a system is composed of structural units called components, each defined by clear boundaries.

A component typically has the following characteristics.

By adopting such a structure, a system becomes resilient to local changes and can evolve incrementally while preserving overall consistency.

As AI-driven code generation becomes widespread, these characteristics become even more important.

AI can rapidly generate the internal implementation of components, but it is not particularly good at autonomously maintaining consistency across component boundaries.

Therefore, in AI-era development, the following division of roles naturally emerges.

CBD becomes the central structural foundation that enables the division of responsibilities between humans and AI in software development in the AI era.

Based on the discussion so far, the development process in the AI era can be organized into the following structure.

The Unified Process (UP) provides the process backbone that defines how development proceeds.

Component-Based Development (CBD), on the other hand, provides the central structural model of development, defining the units in which software is constructed.

AI functions on top of this as an implementation generation engine that produces the internal implementation of components.

From this perspective, the development process in the AI era can be understood as the following three-layer structure.

UP organizes how the project progresses, CBD stabilizes the structure of the system, and AI rapidly generates implementations within that structure.

By separating these roles, it becomes possible to combine rapid AI-driven code generation with the structural stability of the overall system.

This structure also naturally clarifies the division of responsibilities between humans and AI.

Under this division of responsibilities, CBD is positioned not merely as an architectural style but as the central structural model of the development process in the AI era.

Even if CBD becomes the central structural model of development, that alone does not make it possible to build real systems.

To make CBD viable as software, an infrastructure is required that can treat components as executable structural units.

In traditional software development, components were often treated mainly as design concepts, and at the implementation stage they were frequently handled ambiguously as collections of libraries or modules.

However, in the AI era of development, this kind of ambiguity becomes problematic.

To operate AI-generated code in a stable manner, components must be treated as executable structural units with properties such as the following.

Achieving this requires an execution platform centered on components.

CNCF (Cloud Native Component Framework), introduced in this series, is one such execution platform designed to make CBD executable.

In CNCF, a system is organized into the following hierarchical structure.

Through this structure, components are treated not merely as design concepts but as clearly defined structural units that are actually deployed and executed.

With such an execution platform in place, CBD becomes not merely a design philosophy but an operational development model that can be used in real software development.

In this series, we have examined the software development process in the AI era from several perspectives.

First, we introduced the Unified Process (UP) as a guiding framework for thinking about the development process. The principles presented by UP remain important even in the AI era.

On the other hand, in an environment where AI-based code generation becomes widespread, the quality of software is supported not by individual implementation techniques but by the structure and boundaries of the system.

In this article, we therefore organized a basic framework for AI-assisted development by using UP as the backbone of the process and positioning Component-Based Development (CBD) as the central structural model of development.

Within this structure, the following division of roles emerges.

In the AI era of software development, designing the structure of the system becomes more important than writing code itself.

A development process centered on CBD can be considered one of the fundamental frameworks for achieving stable software development through collaboration between humans and AI.