The fundamental question of whether an organization should "build" custom software solutions or "buy" off-the-shelf products to address its operational challenges has been a perennial debate since the inception of the tech industry. This strategic dilemma permeates every level of enterprise, from traditional IT departments grappling with infrastructure needs to agile product teams designing customer-facing applications. While the allure of a ready-made solution often promises speed and cost efficiency, it frequently comes with inherent functional limitations and a degree of vendor lock-in. Conversely, building a bespoke system offers unparalleled customization and competitive differentiation but demands significant upfront investment in time and resources, alongside the perpetual burden of ongoing maintenance and upgrades.
Historically, the decision-making framework has largely hinged on an organization’s core competencies. If a problem directly related to a company’s unique value proposition or competitive advantage, the inclination was to build. For functions outside this core, such as standard administrative tasks or generic business processes, buying a commercial solution was the preferred route. However, this dichotomy has always been an oversimplification. The reality for many enterprises, particularly large corporations, often involves a hybrid approach: purchasing a commercial product and then extensively customizing it to align with specific business workflows, regulatory requirements, or legacy systems. This "buy and tailor" model acknowledges that no single off-the-shelf product perfectly fits every organizational nuance.
For decades, specialized product teams within technology companies enjoyed the luxury of building innovative solutions, while the broader business units typically relied on a centralized IT department to fulfill their extensive and often backlogged list of requests. This created a bottleneck, with IT often perceived as a cost center rather than an enabler of business agility.
The Genesis of User Programming and Democratized Development
The landscape began its gradual transformation with the advent of "user programming," a concept designed to empower non-technical individuals to create software solutions. A pivotal moment in this evolution occurred in 1979 with the introduction of VisiCalc for the Apple II, widely recognized as the first spreadsheet program. VisiCalc enabled business users, primarily accountants and financial analysts, to define complex calculations and models using formulas, effectively "programming" without writing traditional code. This was an enormously empowering development, shifting a portion of application development away from IT specialists and directly into the hands of end-users.
The legacy of VisiCalc is evident today in the countless millions of user-created programs, predominantly formulas and macros, that run daily in spreadsheet applications like Microsoft Excel across virtually every company worldwide. This era further expanded with the release of Visual Basic in 1991, which offered a graphical environment for creating applications, allowing millions more non-developers to build desktop tools and utilities. Visual Basic can be seen as one of the earliest widespread "low-code" options, providing a visual interface and simplified syntax that significantly lowered the barrier to entry for application development. This movement continued to gain momentum, leading to a proliferation of low-code and no-code platforms in the 21st century, designed specifically to enable non-technical users to build functional applications with minimal or no traditional coding expertise, even before the mainstream emergence of generative artificial intelligence.
Generative AI: The New Frontier of User Programming
Today, the landscape of user programming is undergoing another profound transformation, driven by the rapid advancements in Generative AI. This new generation of tools, exemplified by platforms such as Lovable and Bolt, fundamentally redefines the concept of "programming." Instead of requiring specialized syntax or visual blocks, these platforms allow users to express their desired application functionality using natural language, primarily English. This paradigm shift, often referred to as "vibe coding," effectively makes natural language the programming interface, opening up application development capabilities to almost anyone with a problem to solve and the ability to articulate it clearly.
While the "build vs. buy" choice has always existed, and non-technical individuals have often had options to create solutions, the crucial differentiating factors have been the specific skills required to use a particular tool and the types of applications it could realistically construct. The impressive leap offered by contemporary generative AI tools is that the primary skill needed is natural language comprehension and expression, and the scope of applications that can be built is far less constrained than with previous user programming paradigms. This democratizes application creation to an unprecedented degree.
The Enduring Complexity of Enterprise Business Logic
Despite the revolutionary potential of generative AI to empower citizen developers, many industry observers and solution providers are quick to declare the impending demise of Software as a Service (SaaS) vendors, positing that everyone will simply "build" their solutions using AI, rendering "buying" obsolete. However, a deeper analysis reveals why this sweeping prediction is almost certainly premature and overlooks fundamental complexities inherent in enterprise-grade business software.
The core reason why critical business software—spanning procurement, invoicing, payments, budgeting, forecasting, payroll, staffing, sales force automation, customer relationship management (CRM), customer service, and enterprise resource planning (ERP)—is unlikely to be entirely replaced by user-programmed, AI-generated solutions lies in the intricate web of business rules and the associated business logic that underpins these systems.
What many non-technical proponents of pure AI-driven development fail to grasp is that most enterprise business solutions are built upon literally thousands of often complex, interconnected business rules and millions of lines of corresponding business logic. These rules are not arbitrary; they meticulously capture and enforce critical constraints and processes related to policy, regulatory compliance, data security, legal mandates, financial governance, pricing structures, and numerous other operational necessities. For instance, a simple invoicing system must adhere to national tax laws, company-specific discount policies, payment terms, and accounting standards, each represented by layers of complex logic. A procurement system must enforce spending limits, approval hierarchies, vendor compliance, and contractual obligations.
The discernment and codification of these rules represent a monumental effort, often requiring extensive domain expertise, legal review, and collaborative input from various stakeholders. The vast majority of non-technical individuals attempting to create business applications via generative AI tools possess little to no awareness or understanding of these deep-seated business rules. Even experienced technical developers often struggle to fully comprehend these rules, precisely because they are frequently embedded implicitly within existing codebases as business logic. The original architects and implementers of these rules may have long since departed, and while documentation might exist (which is rare for every rule), it seldom provides the crucial context, rationale, and nuanced exceptions behind each rule. This institutional knowledge—the "why" behind the "what"—is precisely what product managers and business analysts historically acquire and internalize to define truly viable and compliant solutions. It is also a significant reason why addressing technical debt in legacy systems is so challenging; one must painstakingly tease out the underlying business rules and determine which remain valid and essential for future operations.
Therefore, building a robust system that can accurately capture, manage, and rigorously enforce thousands of essential business rules, ensuring that all transactions are handled precisely as required by policy and regulation, is an undertaking of considerable complexity and value. This intrinsic complexity forms a formidable barrier to the wholesale replacement of mature SaaS solutions by ad-hoc, AI-generated applications.
The Hybrid Future: "Yes to Both" with Intelligent Integration
However, the enduring value of these rule-driven business solutions does not mean the current SaaS landscape will remain static. While strong SaaS vendors are not facing an existential threat, significant changes are indeed on the horizon. The paradigm is shifting from solutions designed solely for human interaction to those built for interaction with both humans and increasingly sophisticated AI agents, alongside new custom solutions crafted (whether "vibe-coded" or hand-coded) on top of these established component platforms.
This vision of a hybrid future, where enterprises simultaneously "build" and "buy," is enabled by a crucial technological advancement that the industry has long needed: a widely accepted protocol for describing business services in a machine-readable and machine-understandable format. This need became increasingly apparent with the proliferation of the internet and interconnected systems.
Approximately a year ago, Anthropic proposed the Model Context Protocol (MCP), which has rapidly gained traction because it addresses this critical, long-standing architectural challenge. The MCP allows for the semantic description of business services, enabling computers and AI agents to understand the capabilities, inputs, outputs, and constraints of an API or service, rather than merely parsing its syntax. This level of machine comprehension is extraordinarily powerful for complex enterprises.
With the MCP and similar initiatives, the future of enterprise software will see companies continue to procure complex and valuable component services from specialized SaaS vendors for critical parts of their operations. The key difference will be that these bought components will be explicitly designed to be accessed and controlled not only by human users but also by intelligent software agents. Some of these AI agents will likely be developed and provided by the SaaS vendors themselves, offering enhanced automation and proactive capabilities. Others will be created by systems integrators, tailoring solutions for specific enterprise needs. Critically, end-customers will also be empowered to create their own AI agents or generate custom workflows using generative AI tools, which can then orchestrate and interact with these standardized, machine-readable SaaS components.
The Broader Implications of Mainstream User Programming
While user programming has historically operated somewhat on the periphery of enterprise IT, the new generation of AI-powered tools is rapidly bringing these capabilities into the mainstream. This shift is overwhelmingly positive, particularly for the countless individuals and departments that have long been constrained by limited traditional IT resources and lengthy development cycles. It promises greater agility, faster problem-solving for specific departmental needs, and a more empowered workforce.
However, as more non-technical individuals aspire to create solutions beyond simple personal productivity tools, they will inevitably encounter the same fundamental challenges that the professional product development world has grappled with for decades. The most important lesson, often overlooked in the excitement of new tools, is that the hardest part of software development is rarely the actual building or delivering of a solution. The true challenge, and the source of most project failures, lies in discovering the right solution to build—understanding the problem deeply, validating assumptions, identifying critical business rules, and ensuring the proposed solution truly addresses the underlying need without creating new complexities or violating existing constraints.
This will necessitate new forms of governance, training, and collaboration between business users and IT. IT departments will evolve from being sole developers to becoming architects, integrators, and guardians of the enterprise data and security landscape, guiding citizen developers and ensuring their creations align with overarching strategic and compliance frameworks. SaaS vendors, in turn, will need to embrace open standards, robust APIs, and AI-native integration capabilities to thrive in this hybrid ecosystem. The future of enterprise software is not a simple "build or buy" but a sophisticated "build on buy," intelligently orchestrated by humans and AI, all working within a framework of evolving protocols and shared understanding.
