Riding the Fourth Wave of Software Development: How Agentic AI Is Redefining How Software Gets Planned, Built, and Delivered

“Change is the only constant.” – Heraclitus

It’s been 30 years since I first learned to code. At that time, there were no Integrated Development Environments (IDEs), no Agile development, and team-based development was in its infancy. The internet was far from a guaranteed success. Over the past three decades, the only constant has been continuous change—not just in the software we build, but in the way we approach the business process of building and delivering it.

New innovations, languages, and processes have continuously disrupted, challenged, and evolved this relatively young engineering discipline. While mega-trends like object-oriented programming, the rise of the web, globally distributed development, Agile, modern collaboration tools, mobile, and cloud computing have all transformed how we build software, the past three years have introduced perhaps the most disruptive force yet: modern artificial intelligence.

So how will AI change the way we plan, build, test, secure, deliver, run, and support software—now and into the future? To understand where we’re going, we first need to understand where we’ve been. This post (and a series of follow-ups) aims to examine the waves of change in software development, explore why AI is uniquely suited to the software lifecycle, and chart a course for what’s next.

The First Wave: Procedural Development

The first wave of modern software development is often referred to as the age of procedural programming. Developers in this era heavily relied on structured programming techniques, which were highly dependent on loops, conditionals, and sequential logic. Languages like FORTRAN, COBOL, ALGOL, Pascal, and C dominated, and most development teams were co-located—both with each other and their runtime hardware (usually mainframes).

This wave laid the technical foundation for performant, reliable software. But it also came with limitations: top-down decision-making, slow waterfall cycles, and inflexible, hard-to-maintain codebases.

The Second Wave: Object-Oriented Programming, Agile, and the Internet

By the early 1990s, the software development landscape had begun to shift. The internet was evolving from a hobbyist’s playground into the backbone of the modern economy, creating massive pressure to modernize, scale, and accelerate software delivery.

This wave introduced object-oriented programming, with languages like C++, Java, Smalltalk, Python, and C#, enabling abstraction, encapsulation, inheritance, and laying the foundation for code reuse and maintainability.

Equally as important were changes to process. Agile development, globally distributed teams, pair programming, and leaner planning methodologies gained traction. Tooling followed quickly: IDEs, source control management (SCM), and build automation became standard. Agile planning, test automation, delivery pipelines, and continuous integration tools went from “nice-to-haves” to essentials.

This combination of internet-scale opportunity and improved tooling led to an explosion of productivity; many estimate that software output grew five to ten times in the 1990s alone, compared to all previous decades combined.

While this marked a huge improvement, real change and innovation were just beginning.

The Third Wave: Mobile, Cloud, SaaS, and Modern DevOps

As we crossed into the 21st century, the software industry entered a period of exponential growth. High-speed internet, more powerful personal computers, smartphones, massive data centers, and cloud computing catalyzed the Third Wave of software development.

What set this wave apart went beyond new programming languages. The Third Wave was defined by ubiquitous connectivity, scalable computing power, and a market environment where every company was becoming a software company, regardless of industry.

This wave brought deeper levels of automation, user-centric design, and “consumerized” software experiences. Agile methodologies evolved into DevOps, emphasizing the seamless integration of development and operations. Concepts like CI/CD became the norm. Virtualization and cloud computing made infrastructure elastic and instantly scalable.

Business models also changed: SaaS, platform ecosystems, and the API economy emerged. The software industry expanded rapidly, with startups, scale-ups, and incumbents racing to transform. As Marc Andreessen famously said in a 2011 Wall Street Journal op-ed: “Software is eating the world.”

The Fourth Wave: Agentic Software Development and Delivery

While AI, machine learning, and neural networks have been around since the mid-20th century, November 2022 marked a major inflection point. That’s when OpenAI released ChatGPT 3.5, bringing generative AI large language models (LLMs) into the mainstream. The results were immediate and profound. ChatGPT showcased human-like context understanding, fluid natural language interactions, and multi-modal capabilities. With these breakthroughs, agentic AI—AI that can perceive, reason, and act with autonomy—is no longer science fiction. It’s here, and it’s changing everything.

While AI will impact every business function, its influence is most visible in software development. We’re now in the early days of the Fourth Wave—driven by AI and agentic workflows. This wave is about far more than just auto-complete or code generation. AI agents are reshaping how we plan, write, test, secure, deploy, run, and support software. They’re optimizing each link in the chain. Even more importantly, they’re changing the entire system.

In large organizations, we’re already seeing a pattern: narrow, siloed AI adoption yields only incremental gains. Automating a single subprocess shifts bottlenecks rather than eliminating them. The real value emerges when teams adopt AI holistically, treating the software lifecycle as an interconnected system.

A combination of robust, end-to-end understanding of your development and delivery pipeline with thoughtful, strategic AI adoption will separate the winners from the also-rans in this new era.

Throughout this series, we’ll dive into the core attributes of the Fourth Wave and explore real-world examples of successful (and unsuccessful) AI adoption in software teams, and outline frameworks for getting it right.

What’s Next?

Software development and delivery is a natural fit for AI, yet right now, AI in software is being deployed where it’s easiest — and underused where it matters most.

Nearly half of all venture capital in the AI-for-dev space is going to code assistants that help developers write code faster. Instead, we need to be thinking deeply about what happens before and after that moment: AI in planning, AI in delivery, AI in quality assurance, AI in security. That’s where the opportunity is.