Why It's Time to Rethink How We Build Enterprise Software

Introduction

Artificial Intelligence has been part of the conversation at Falcon for some time. More importantly, it has already become part of the way we build software.

More recently, that conversation has taken on greater relevance within our Dynamics practice and across the wider Microsoft Dynamics community. In discussions with colleagues, customers and friends, I found the same questions coming up again and again.

Some were excited about Artificial Intelligence. Others were skeptical. To be honest, for quite some time I found myself somewhere in the middle.

After spending the past months experimenting with LLM-assisted engineering in my own Dynamics work and challenging many of my assumptions, I reached a simple conclusion.

Artificial Intelligence is no longer something we should observe from a distance. It is becoming part of the way enterprise software is engineered.

That is why I decided to write this series.

Every Major Platform Shift Starts with Skepticism

Our journey at Falcon started in the days of Microsoft Axapta 2.5 and 3.0. Since then, we have experienced every major evolution of Microsoft's business applications—from Dynamics AX to today's Dynamics 365 Finance & Operations and Business Central.

Every major transition forced us to rethink how we designed solutions, how we developed software and, sometimes, even how we organized our teams. Eventually, every one of those changes became the new normal.

I believe Artificial Intelligence represents another one of those moments.

What Changed My Perspective on AI in ERP

Like many ERP professionals, I approached AI with a healthy dose of skepticism.

Not because I doubted the technology itself, but because enterprise software is different. In ERP, every customization has business consequences, and understanding the business is just as important as writing the code.

For quite some time, I wasn't looking for reasons to adopt AI. I was looking for reasons why it wouldn't work in Dynamics projects.

At some point, I realized that no article, conference or online debate was going to answer my questions. The only way to understand AI was to start using it myself.

So I did.

I started experimenting with AI in my own engineering work, gradually integrating it into real Dynamics 365 Finance & Operations development.

What I discovered wasn't that AI could replace engineers. It was that the better the business context I provided, the more valuable AI became.

That was the moment I realized I had been asking the wrong question.

The question was never whether AI could replace ERP engineers.

The real question is whether we, as ERP engineers, are willing to evolve the way we build software.

We Didn't Start Yesterday

Although our focused exploration of AI within Dynamics has accelerated over the past months, we are not starting from scratch. It builds on decades of experience designing, extending and maintaining business-critical ERP solutions.

Within Dynamics, the questions are different. Can large language models work effectively with ERP context? Can they help engineers navigate X++ and AL codebases, application metadata, legacy customizations and platform-specific constraints? Can they improve productivity without compromising quality, maintainability or accountability?

Over the past months, we have been testing these capabilities directly within our Dynamics engineering workflows. What we discovered is that AI can help engineers understand existing implementations, generate and review code, automate repetitive tasks and work more efficiently.

But the value does not come from the model alone. It depends on the context provided to it and, above all, on the expertise of the engineer using it. AI does not understand a business better than an experienced consultant or developer; it amplifies the capabilities of those who already do.

Every organization has its own processes, constraints and goals, and translating them into reliable enterprise software still requires deep business expertise. In finance, manufacturing, warehousing, logistics or procurement, that understanding is often the difference between software that technically works and software that actually supports the business.

That expertise remains the foundation of every successful enterprise solution.

A Message to the Community

The future is not about replacing software engineers. It is about redefining what great software engineering looks like.

I understand why many ERP engineers remain skeptical because, until recently, I shared many of the same questions.

Today, I believe the engineers who continue learning and combine business expertise with AI-assisted engineering will help build the next generation of enterprise software.

Building the Future Together

By combining decades of Microsoft Business Applications experience with the knowledge already built by our AI engineering team, we have accelerated our own learning.

Over the past months, we have also been developing an internal agent harness that is evolving into the foundation of our Falcon AI Engineering Framework.

In the next articles of this series, I will share the adoption challenges we are encountering, the lessons we are learning and the engineering approach we are building around AI.

Looking Ahead

We don't know exactly what enterprise software engineering will look like five years from now. None of us do.

What we do know is that evolution has always been part of our journey. From Axapta 3.0 to today’s cloud-based business applications, we have continued to learn, adapt and build.

AI is not the reason we started evolving. It is the next chapter in that evolution.

As we have done through every major platform shift, we will keep learning, experimenting and building.

I hope you'll join us on that journey.

AI-Driven Enterprise Solutions - Part 1: Why It's Time to Rethink How We Build Enterprise Software