For a while now, much of the conversation around AI has been focused on efficiency and cost reduction. This is largely because businesses are looking for quick wins and there's plenty of low-hanging fruit to be found streamlining workflows and automating routine tasks. After all, these initiatives tend to produce quick, visible results that build confidence and demonstrate that AI can deliver real value.
But, in the broader journey of AI adoption, efficiency is only the starting point. Once those early gains are realized, a more important question emerges: how can AI make your business harder to compete with? Because while automation improves how you operate, strategic application of AI can strengthen how you differentiate.
In this article, we'll introduce a simple way to rethink your software portfolio and organize it around strategic impact. We'll examine why differentiation often goes underfunded, how AI changes the economics of investing in the enhancement of complex competitive systems, and why the greatest opportunity isn't funding that routine ERP upgrade project but strengthening the part of your business that competitors can't replicate.
Rethinking Your Software Portfolio
Every business operates on a portfolio of software systems. Some of these systems exist to keep the lights on. Others support internal coordination and day-to-day operations. A smaller set directly influences how you price, produce, serve customers, manage risk, or allocate resources. These systems shape your competitive position in the market. Yet over time, most technology investments gravitate toward modernization, compliance, and incremental efficiency gains. While necessary, those investments rarely strengthen the capabilities that truly differentiate a company from its competitors.
In his book, Domain-Driven Design, Eric Evans famously observed that not every part of a business contributes equally to competitive advantage. Based on this, he organized business capabilities into three groups, a distinction that becomes especially useful when evaluating where your systems and investments should focus.
Core: This is where your differentiation lives. Depending on your industry, we could be talking about your unique just-in-time logistics model, production optimization logic, service delivery, risk methodology, or the way you configure complex products. These capabilities are nuanced, experience-driven, and often difficult for competitors to replicate. This is where margin is protected and advantage is created.
Supporting: These capabilities enable the business to operate effectively, but they do not define your uniqueness. They include internal workflows, reporting tools, and coordination systems that help teams execute consistently and efficiently.
Generic: These are essential but commoditized functions such as payroll, email, HR platforms, and infrastructure. Every company relies on them, but nobody gains strategic separation because of them.
When we apply this lens to your software portfolio, a revealing pattern often emerges. As you can see in Figure 1 below, a significant portion of your software investment and attention goes toward generic and supporting systems. These systems must be stable and efficient, so they naturally command budget and oversight. Meanwhile, the systems that directly enable differentiation often remain fragmented, manually driven, or under-modeled because they're more complex and difficult to standardize.
Figure 1: Where Complexity and Differentiation Converge
Rethinking your software portfolio means organizing it around strategic impact instead of historical purchasing decisions. It means asking three simple questions:
Which systems merely keep us operational?
Which systems support coordination and efficiency?
Which systems directly strengthen our competitive position?
Once those distinctions are clear, investment decisions become more intentional. Generic systems should be reliable and cost-effective. Supporting systems should enable operational excellence. Core systems, however, deserve disproportionate attention because they shape how you win.
Why Differentiation Gets Neglected
One of the great paradoxes of software investment strategy is that the systems that drive the most competitive impact are usually the least likely to be targeted for meaningful upgrades or enhancements. As counterintuitive as this may seem, there are actually many practical reasons for this:
Perceived Operational Risk: Core systems are deeply embedded in revenue-generating processes, so leaders fear disruption more than they value potential upside.
High Complexity and Ambiguity: The business logic in these systems is nuanced, exception-heavy, and often poorly documented, making upgrades feel unpredictable and difficult to scope.
Dependencies on Institutional Knowledge: Critical decision logic frequently lives in experienced employees rather than in clearly defined systems, which makes modernization feel dangerous.
The “If It Ain’t Broken” Mentality: If the system isn’t visibly failing and it's working "well enough", it’s hard to justify the disruption of changing it.
For these reasons (and others), the perception is that there's a high level of risk associated with touching core systems. Gaps in supporting and generic systems, by contrast, are much easier to address. For example, workflow automation initiatives produce measurable efficiency gains. Similarly, infrastructure improvements increase security and reliability. These projects are easier to justify because the risks are clearer and the outcomes are more predictable.
Figure 2: A Classic Example of If it Ain't Broke...Don't Fix it
The result is a predictable pattern. Investment flows toward stability, compliance, and incremental efficiency improvements, while the systems that directly shape competitive positioning receive modest enhancements rather than deliberate strategic focus.
This isn't a failure of leadership; it's a rational response to perceived risk. But over time, that steady bias toward safer investments introduces another form of risk as the distance between you and your competitors narrows. In other words, differentiation erodes gradually when it is not intentionally strengthened.
AI Changes the Economics of Complexity
Historically, enhancing core systems required exhaustive requirements gathering, rigid specifications, and long development cycles. In financial terms, we're talking about some huge capital projects. The more nuanced the business logic, the more fragile and expensive the effort became. In that environment, complexity was something to avoid, not embrace.
However, the rise of AI technology has completely changed this investment calculus. Let's take a look at some reasons why.
Decoding the Undocumented
One of the biggest obstacles to enhancing legacy systems is gaining a clear understanding of how they actually work. Over the years, documentation tends to become outdated and the original architects are probably long since gone or even retired. Regardless, critical logic is probably living partly in code and partly in the heads of experienced employees.
Before we can make any meaningful progress, we must first understand how these systems were originally constructed. This can be a very long and arduous task, especially if the legacy systems have accumulated multiple layers of customizations, patches, workarounds, and one-off enhancements.
Fortunately, these are problems that AI-powered large language models (LLMs) are particularly good at solving. For perspective, have a look at Figure 3 below. Here, we're utilizing GitHub Copilot to examine some Python code to understand what it's doing. These kinds of tools can really accelerate the time it takes to come up to speed with a legacy code base, and they're getting better every day. Tools like Claude Code and Cursor are scaling to support more complex documentation projects that deeply analyze an entire code repository.
Figure 3: Using GitHub Copilot to Interpret What Legacy Code is Doing
AI as a Strategic Thought Partner
Once we get to a place of understanding, the next step in the flow is even more daunting. From here, the conversation shifts from What does this system do? to What could it become?
This is where more general-purpose LLMs begin to play a different role. They're not just code assistants. They've been trained on a vast body of technical, operational, and industry knowledge. When grounded in your domain context, your data, and your business constraints, they can also function as strategic collaborators.
Figure 4: Working with the Researcher Agent in M365 Copilot
Instead of starting with a blank whiteboard, we can begin with structured exploration. We can ask:
How might we redesign this operating model to improve margin under different market conditions?
What optimization techniques are common in adjacent industries that we have not applied here?
Where could automation augment expert judgment without replacing it?
What alternative workflows could reduce friction in our most complex processes?
AI can generate scenarios, compare approaches, surface tradeoffs, and synthesize best practices from across industries. It can also help frame opportunities, stress-test assumptions, and refine concepts before significant capital is committed. Ideas that once required weeks of cross-functional workshops can be explored in hours, with sharper questions and more informed starting points. In this context, AI becomes a thought partner that expands the design space, helps articulate possibilities, and sharpens proposals before they move into development.
Accelerating the Path from Idea to Implementation
Once the business has clarity on where it wants to strengthen its competitive edge, the next barrier is execution. AI-powered development tools compress that timeline dramatically.
As we observed earlier, modern AI assistants like GitHub Copilot do far more than autocomplete code. They help translate business intent into technical implementation. A product owner can describe the desired behavior of a pricing engine or scheduling optimizer in plain language, and the development team can use AI to rapidly generate initial scaffolding, draft algorithms, and propose implementation patterns. Indeed, it's truly incredible how fast we can go from initial idea to working prototype these days.
Figure 5: Rapid App Development Using Claude Code
In addition to increased development velocity, AI also changes how teams collaborate around complex logic. Instead of manually drafting and refining intricate rule sets, developers can iterate with AI over decision trees, constraint models, and optimization algorithms. They can simulate scenarios, compare alternative approaches, and refine edge cases in minutes rather than days. Developer tools like Claude Code become a co-innovator, helping surface options that may not have been immediately obvious.
This is especially powerful when surgically enhancing core systems where the business logic is rarely straightforward. Pricing rules interact with customer segments. Production schedules must balance capacity, lead times, and supply variability. Risk models incorporate multiple data sources and probabilistic assumptions. AI-assisted development allows teams to explore and refine these interactions more fluidly, testing hypotheses and adjusting logic before committing to large-scale changes.
Before moving on, we should also mention how AI supports the entire development lifecycle. It can generate unit test cases, refactor legacy components, suggest performance improvements, and document changes automatically. This reduces the mechanical overhead that often slows progress, allowing engineers to focus more time on higher-order design decisions and domain refinement.
Making Change Safer and More Affordable
Any time core systems are involved, change needs to be approached carefully. Before releasing changes, teams need confidence that new functionality works as expected and that existing behaviors continue to operate reliably.
Historically, maintaining that level of confidence has required significant effort. Regression test suites can grow large and complex, and writing, maintaining, and executing those tests takes time. Manual testing is often used to fill in the gaps, but it can be difficult to sustain as systems evolve and development cycles become shorter.
AI-powered testing tools are helping make this process more manageable. Platforms such as Tricentis Tosca use AI to assist with generating and maintaining automated tests. They can help identify fragile tests, adapt to minor interface changes, and suggest ways to improve overall test coverage. By learning from application behavior over time, these tools can help teams maintain more reliable testing processes with less manual effort.
Figure 6: Coordinating Complex Regression Tests Using Tricentis Tosca
Differentiation is Where Profit Lives
As we observed in the previous section, AI technology has drastically changed the risk-reward ratio. And as the cost and risk of working with complex systems continues to come down, you have a real opportunity to rethink where you focus your innovation efforts. Instead of limiting improvements to operational efficiency, you can start looking more closely at the capabilities that directly shape their competitive advantage.
This is where the biggest opportunities tend to live.
The core systems that fuel your competitive advantage often have an outsized impact on performance. Small improvements in these areas can translate into meaningful gains in margin, responsiveness, and customer experience. Unlike generic systems, where improvements mostly drive efficiency, enhancements to core capabilities can strengthen how your company competes.
In the past, many of these initiatives would have been dismissed as too risky or too difficult to implement. Today, AI-powered discovery, development, and testing tools make it much easier to explore these opportunities with smaller, more controlled investments.
That shift opens the door to a different kind of innovation. Instead of playing it safe around the edges, you can begin pushing the envelope where it matters most. And if those calculated bets pay off, the impact can be wildly transformative.
Institutionalizing Your Edge
As we've seen, a lot of technical debt exists within most, if not all of your core systems. Leaders are usually aware this debt exists, but because these systems continue to run, the risk often gets treated as manageable—something to live with until there is a compelling reason to intervene.
That mindset creates a subtle illusion: the idea that technical debt remains stable as long as the system is left alone.
In reality, the debt rarely stays static. It grows quietly over time. Team members move on. Original architects retire or change roles. The unwritten context that explains why certain design decisions were made slowly fades. What remains is a system that still works, but whose inner logic becomes increasingly fragmented and harder to interpret.
This erosion of institutional knowledge can be just as risky as the code itself. Even if no new features are added, the organization’s ability to confidently manage and improve the system declines. Eventually, the technical debt ceiling rises to a point where meaningful change feels even more intimidating than before.
This is where the AI tools discussed throughout this article offer a practical opportunity. Even if the timing isn't right for a major innovation initiative, you can at least start capturing and organizing the knowledge that surrounds your core systems. AI can help analyze codebases, explain undocumented logic, summarize complex workflows, and translate technical implementation into plain-language documentation.
In effect, these tools allow you to “download” and preserve institutional knowledge that might otherwise be lost.
By strengthening documentation and clarifying how critical systems actually work, you can establish a stronger foundation for whatever comes next. Future enhancements become easier to plan. Risks become easier to evaluate. And the next generation of engineers and business leaders inherit systems that are more understandable and more manageable.
Closing Thoughts
For a long time, the complexity of core systems discouraged meaningful investment. Even when leaders recognized the potential value locked inside those capabilities, the perceived cost and risk of engaging with them made safer projects easier to justify. As a result, many organizations focused their technology efforts on areas that improved efficiency while leaving the systems that shape competitive advantage largely untouched.
AI is beginning to change that dynamic. By lowering the cost of understanding, enhancing, and safeguarding complex systems, it opens the door to more intentional investment in the capabilities that define how a business competes. Whether the next step is strategic innovation or simply capturing the institutional knowledge embedded in legacy platforms, you now have new tools to strengthen and evolve the systems at the heart of your competitive edge.
