Why Software is Not Enough

There’s a story Michael Bloomberg shares in his memoir “Bloomberg by Bloomberg” that doesn’t get told often enough. Sometime in the late 1970s or early 1980s, fresh from his departure from Salomon Brothers, Bloomberg encountered Theodore Levitt’s seminal essay “Marketing Myopia.” The piece, written two decades earlier, contained a deceptively simple insight that would reshape how Bloomberg thought about building his nascent financial information company. Levitt’s argument was this: railroad companies failed not because people stopped needing transportation, but because the executives defined themselves too narrowly. They thought they were in the railroad business when they were actually in the transportation business. The product—trains—blinded them to the broader capability they possessed.

Bloomberg internalized this lesson with the kind of clarity that separates builders from operators. He understood that he wasn’t creating a terminal or even software. He was building a capability for financial decision-making at scale, a nervous system for markets that happened to manifest through hardware and code. The Bloomberg Terminal became legendary not because of its amber text on black screens, but because it represented expertise, access, and analytical power that transcended any single technological implementation.

The New Myopia: Mistaking Software for Strategy

Now we find ourselves at a peculiar moment in history where that same lesson needs to be relearned, but the stakes are vastly higher and the myopia more seductive. Today’s myopia isn’t about trains versus transportation. It’s about software versus business capability. And just as the railroad executives couldn’t see beyond the steel rails and steam engines, today’s leaders are mesmerized by their codebases, their deployments, their elegant microservices architectures. They’ve convinced themselves they’re building software when they’re actually building businesses—and that distinction is about to matter more than it ever has.

Consider what’s actually happening beneath the surface of our economy right now. Software, for all its revolutionary impact over the past fifty years, operates on a fundamentally static model. You identify a problem, you write code, you deploy a system, and then—critically—that system remains relatively fixed until the next major release cycle. It’s better than paper and filing cabinets, certainly, but it’s still a snapshot solution to a dynamic problem. The software reflects business logic as it was understood at the moment of creation, encoding yesterday’s assumptions into tomorrow’s operations.

This worked reasonably well in a world where business models evolved on the timescale of years or decades. But we no longer inhabit that world. The pace of business model innovation has accelerated to the point where traditional software development cycles look less like enablement and more like anchors dragging behind organizations struggling to adapt. By the time you’ve deployed your new system, the competitive landscape has shifted, customer expectations have evolved, and the very value chain you optimized for has been disrupted by someone who didn’t bother playing by the old rules.

Three Converging Shifts That Change Everything

Enter artificial intelligence, which represents not merely an incremental improvement in computational power but a phase transition in what’s possible at the intersection of business capability and technological implementation. Three fundamental shifts are converging simultaneously, and together they render the traditional software paradigm not just outdated but actively dangerous to organizational survival.

The Economics of Augmented Work

The first shift is the most obvious but least understood: the workforce itself can now be augmented and automated through AI in ways that fundamentally alter the economics of every business model. I’m not talking about simple robotic process automation or the replacement of call center workers with chatbots. I’m talking about the ability to take knowledge work—the analysis, synthesis, decision-making, and creative problem-solving that has always required human cognition—and amplify it by orders of magnitude. The workforce line item, typically the largest cost in any service business, is about to become radically more productive while simultaneously more flexible. This isn’t a marginal improvement. This is the equivalent of discovering that your most constrained resource is suddenly abundant.

From Static Systems to Dynamic Neocortex

The second shift is subtler but more profound: the very nature of software systems is transforming from static deployments to dynamic, bespoke enablement layers. Think of it as the difference between building a road and having access to a transportation network that reconfigures itself in real-time based on where people need to go. Traditional software encoded business logic in rigid structures. The emerging paradigm creates what I think of as a neocortex layer—an intelligent membrane that spans across business domains, learning and adapting continuously rather than waiting for the next release cycle. This neocortex doesn’t replace your operations; it augments them with real-time intelligence that can sense patterns, anticipate needs, and orchestrate resources in ways that were simply impossible when everything had to be pre-programmed.

The Possibility of First Principles Reimagination

The third shift is methodological: we now have the capability to reimagine business models not by tweaking existing processes but by fundamentally rethinking the entire value chain from first principles. This is where the Bloomberg lesson becomes urgently relevant again. Just as he understood that terminals were merely the instantiation of financial decision-making capability, we must understand that our current business operations are merely one possible instantiation of the underlying capabilities we possess. The railroad executives couldn’t see beyond trains. Are we unable to see beyond our current workflows?

The Four-Step Methodology for Business Re-engineering

But here’s where most organizations stumble. They take these powerful new capabilities and apply them to their existing business model, using AI to make their current processes incrementally more efficient. This is the equivalent of using the internal combustion engine to build faster horse-drawn carriages. It misses the entire point. The opportunity—and the necessity—is to fundamentally reimagine what business you’re actually in, just as Bloomberg did when he read Levitt’s essay.

Let me walk you through what this reimagining actually looks like in practice, because the methodology matters enormously. This isn’t about sprinkling AI pixie dust on your org chart and hoping for transformation. It’s a deliberate, systematic process that requires both intellectual rigor and organizational courage.

Step One: Question Everything About Your Business Model

Step one is business model rethinking from first principles. This means genuinely questioning what capability you provide to the world, independent of how you currently deliver it. If you’re a bank, are you in the lending business or the risk assessment and capital allocation business? If you’re a hospital, are you in the healthcare delivery business or the health outcome optimization business? If you’re a logistics company, are you in the shipping business or the supply chain certainty business? These distinctions aren’t semantic games. They determine what’s possible and what’s myopic.

You must be willing to look at your organization with the same fresh eyes that an outsider would bring, stripping away the accumulated barnacles of “how we’ve always done things” to expose the core value you actually create.

Step Two: Find the Disintermediation Opportunities

Step two is the decomposition of your value chain with a specific goal in mind: understanding where disintermediation opportunities exist. Every business is a series of connected activities, and many of those connections exist for historical reasons rather than fundamental ones. They’re there because of limitations in coordination, information flow, trust, or transaction costs—limitations that AI can often eliminate entirely. Map out every step in how value flows through your organization, from initial customer contact through delivery and support.

Then ask the uncomfortable question: which of these steps exist only because we lacked the capability to do things differently? Which intermediary layers, which handoffs, which coordination meetings and approval processes are actually workarounds for constraints that no longer exist?

Step Three: Build Your Capability Ontology

Step three is crafting a full business capability ontology. This is where most organizations reveal their true level of self-understanding. You need to create a comprehensive map of what your organization can actually do—not what software systems you have, but what capabilities you possess. Can you assess credit risk? Can you negotiate with suppliers? Can you diagnose medical conditions? Can you design supply chain routes? Can you manage regulatory compliance?

Each of these capabilities has both a knowledge component and an execution component, and many of them have traditionally been embedded in the heads of your most experienced people. The ontology is about making that tacit knowledge explicit and structured so that it can be augmented rather than merely preserved.

Step Four: Design Your Bespoke AI Neocortex

Step four is where the previous three steps converge into something genuinely new: you develop and design a bespoke AI neocortex to augment the remaining human-centered capabilities at scale. Notice I said “remaining” human-centered capabilities. By the time you reach step four, you’ve already identified which activities can be automated entirely and which require the judgment, creativity, and contextual understanding that humans uniquely provide. The neocortex isn’t about replacing humans. It’s about creating an intelligent layer that makes every human in your organization vastly more capable by providing them with real-time analysis, predictive insights, automated routine tasks, and decision support that adapts to their specific context and needs.

Understanding the Neocortex Architecture

This neocortex concept is crucial to understand because it represents a fundamentally different architecture than traditional software. Traditional software says: here’s how we do accounts receivable, encoded in a system that everyone must follow. The neocortex says: here are the principles and goals of accounts receivable, and I’ll help you achieve them in whatever way makes sense for this particular situation, learning from every interaction to become more effective over time. One is rigid scaffolding. The other is an exoskeleton that amplifies capability while preserving flexibility.

The Asymmetric Advantage of Getting This Right

The implications of getting this right versus getting it wrong are not incremental. They’re existential. Organizations that continue to think in terms of software deployments will find themselves competing against organizations that think in terms of dynamic business capabilities. It’s not a fair fight. It’s the railroad companies competing against airlines while still insisting that the future of transportation involves better rail schedules.

But here’s the deeper truth that makes this moment both terrifying and exhilarating: the methodology I’ve outlined isn’t a playbook for incremental improvement. It’s a framework for wholesale business model reinvention. And business model reinvention, when done properly, creates asymmetric opportunities. The organizations that embrace this approach early won’t just be slightly better than their competitors. They’ll be operating in a different category entirely, with cost structures, speed of adaptation, and customer value propositions that simply cannot be matched by organizations still trapped in the old paradigm.

The Questions That Matter Now

So when someone tells you they’re building software, the right response is to ask: what business capability are you actually enabling? When someone shows you their digital transformation roadmap, the right question is: have you reimagined your business model or are you just automating your current inefficiencies? When someone pitches you their AI strategy, the right challenge is: is this a neocortex that amplifies human capability or just another static system that will be obsolete before it’s fully deployed?

Bloomberg got it right fifty years ago when he understood that terminals were just the instantiation of something larger. We need to get it right now, understanding that software is just one possible instantiation of business capability—and in the age of AI, probably not even the most important one. The question isn’t whether you’re going to build better software. The question is whether you’re going to build better businesses. And that requires seeing beyond the code to the capabilities that actually matter.

The myopia of software is the myopia of our era. The cure is the same one Levitt prescribed in 1960: define your business by the capabilities you provide and the problems you solve, not by the products you happen to ship today. Everything else is implementation details. And in a world of rapidly evolving AI capabilities, implementation details change faster than you can deploy them.

Opinions my own.

Angel Armendariz