For decades, a fundamental strategic question has echoed through the corridors of technology companies and IT departments worldwide: should an organization "build" custom software solutions to address specific problems, or "buy" off-the-shelf products? This perennial dilemma, deeply ingrained in the tech industry’s DNA, has influenced countless decisions in traditional IT infrastructure and modern product development teams alike. While the core considerations of cost, time, and functionality have long driven this choice, the advent of new technologies, particularly Generative Artificial Intelligence (AI), is now reshaping the landscape, introducing unprecedented capabilities and complex implications for the future of enterprise software.
The Enduring Strategic Calculus of Build vs. Buy
Historically, the decision to build or buy has been multifaceted. Buying often presents a faster deployment time and lower initial cost, leveraging a vendor’s existing expertise and infrastructure. However, purchased solutions frequently come with functional limitations, requiring compromises or necessitating costly customizations to align with specific business needs. This is particularly true for large enterprises where unique workflows and legacy systems often demand tailored integrations. Conversely, building a custom solution offers unparalleled control, allowing for precise alignment with core competencies and competitive differentiation. Yet, it entails significant upfront investment in development, ongoing maintenance, and the inherent risks associated with complex software projects, including potential delays and budget overruns.
For most organizations, the prevailing wisdom has been to build solutions that represent a core competency—something integral to their unique value proposition or competitive advantage. Conversely, non-core functions, such as standard HR, finance, or customer relationship management (CRM) tasks, were typically outsourced or addressed with commercial off-the-shelf (COTS) software. This distinction, however, has always had its exceptions, particularly for highly specialized problems where no viable "buy" option exists. The reality, especially in large corporations, often blurred the lines, with a common hybrid approach involving the purchase of a commercial product followed by extensive customization to meet specific enterprise requirements. This dynamic frequently led to a backlog of requests for internal IT departments, as business units, historically lacking the tools to create their own solutions, relied heavily on central IT to implement and adapt commercial software or develop bespoke applications.
A Historical Arc: The Rise of User Programming
The ability for non-technical individuals to create software solutions, often referred to as "user programming," has a longer history than many realize, gradually eroding the exclusive domain of professional developers. This evolution marks a significant democratization of technology, empowering business users to address their specific challenges without constant reliance on specialized IT resources.
The first major breakthrough arrived in 1979 with VisiCalc, the pioneering spreadsheet program designed for the Apple II. VisiCalc was revolutionary not merely as a piece of software, but as the first widespread "user-programming tool." It allowed non-technical business professionals to create complex financial models, budgets, and forecasts using formulas – a form of programming – without writing a single line of traditional code. This was enormously empowering, enabling rapid data manipulation and analysis, and effectively kickstarting the personal computer revolution in business. Its success underscored a deep-seated need for tools that allowed domain experts to directly apply their knowledge to computational problems.
Two decades later, Visual Basic (VB), first released by Microsoft in 1991, further expanded the horizons of user programming. While requiring a more structured approach than spreadsheets, VB offered a graphical development environment that significantly lowered the barrier to entry for creating Windows applications. Millions of "citizen developers" leveraged VB to build custom tools, utilities, and even departmental applications, making it arguably the first widely adopted "low-code" option. Its drag-and-drop interface and event-driven programming model allowed non-professional developers to visually design user interfaces and then attach code snippets, simplifying the complex process of application development.
The momentum continued with the broader low-code and no-code (LCNC) movement, which gained significant traction in the 2010s and continues to grow. Platforms from vendors like OutSystems, Mendix, Bubble, and Microsoft’s Power Apps provide visual development environments, pre-built components, and intuitive interfaces that enable business users (often called "citizen developers") to build sophisticated applications, automate workflows, and create data visualizations with minimal or no traditional coding. Industry analysts have projected substantial growth in this sector, with Gartner forecasting that low-code application platforms will account for over 65% of all application development activity by 2024. This wave has empowered businesses to accelerate digital transformation, reduce IT backlogs, and foster innovation from within.
Generative AI: A New Frontier for User Creation
While the LCNC movement laid a strong foundation, the emergence of Generative AI is now ushering in a transformative new generation of user-programming tools. Products such as Lovable and Bolt exemplify this paradigm shift, where the "programming language" itself has evolved from structured code or visual blocks into natural language – primarily English. This development is profoundly significant because it opens the capability to create complex applications to virtually anyone who can articulate a problem in plain language. The skill barrier has plummeted, making application development accessible to an unprecedented segment of the workforce.
These GenAI-powered tools leverage large language models (LLMs) to interpret user prompts and generate code, configurations, or even entire application structures. This process, sometimes termed "vibe coding," allows users to describe their desired functionality, user interface, or data interactions, and the AI translates these natural language instructions into a working solution. The implications are vast: faster prototyping, reduced development cycles, and the potential for a massive surge in custom applications tailored to highly specific, often niche, business needs that previously would have been too costly or time-consuming to address.
However, it is crucial to clarify that the core choice between building and buying, and the availability of options for non-technical people, has always existed. The key variables have consistently been the specific skills required to utilize a given tool and the types of applications it was designed to build. What makes the new generation of GenAI user-programming tools so impressive is the dramatic reduction in the requisite skill set – primarily natural language proficiency – and the significantly expanded scope of application types that can now be created. This promises to bring user programming from the margins into the mainstream of business operations.
The Nuance of Enterprise Software: Beyond Simple Solutions
Despite the revolutionary potential of Generative AI in democratizing software creation, a prevailing misconception suggests that this new wave will render traditional enterprise software and SaaS vendors obsolete, leading to a future where "everyone will build, and nobody will buy." This perspective, however, largely oversimplifies the intricate realities of modern business operations. The robust SaaS players, integral to the global economy, are far from doomed.
The fundamental reason why critical business software—spanning procurement, invoicing, payments, budgeting, forecasting, payroll, staffing, sales force automation, customer relationship management, and customer service—is unlikely to be entirely replaced by user-programmed solutions lies in the inherent complexity of business rules and the associated business logic. Most people, including many technical professionals, underestimate the sheer volume and intricacy of these underlying components. Enterprise business solutions are underpinned by literally thousands of often-complex business rules and millions of lines of business logic. These rules and the code that implements them encapsulate critical constraints and processes related to policy, regulatory compliance, security protocols, legal mandates, financial regulations, pricing structures, and much more.
The process of discerning, codifying, and maintaining these rules is a monumental undertaking. It requires considerable time, deep domain expertise, and meticulous effort. The vast majority of non-technical individuals aspiring to create business applications using GenAI tools possess little to no awareness of these deeply embedded business rules. Even experienced technical personnel frequently struggle to fully grasp these rules because they are often dispersed throughout legacy codebases, with the original architects and implementers having long since moved on. While documentation might occasionally exist, it rarely provides the granular detail, rationale, and nuanced context behind each rule.
This intricate knowledge forms the bedrock of viable solution definition, a task traditionally undertaken by product managers and, before them, business analysts. It is also why addressing technical debt in enterprise systems is so challenging; it necessitates a painstaking process of extracting these embedded business rules, evaluating their continued relevance, and deciding which ones still apply in evolving operational contexts. A system capable of accurately capturing, managing, and enforcing these thousands of essential business rules, ensuring that transactions are handled precisely as required, represents an enormous investment but yields immense value. This critical layer of complexity is not easily replicated by ad-hoc, user-generated applications, regardless of how sophisticated the underlying AI tools become.
The Future: A Hybrid Ecosystem Powered by AI and Protocols
While strong SaaS vendors are not poised to disappear, the landscape of enterprise software is undoubtedly on the cusp of significant transformation. The traditional model, where business solutions were primarily designed for human interaction, is evolving. Going forward, these solutions will increasingly be utilized not only by human users but also by AI agents and new custom solutions, whether "vibe-coded" by business users or hand-coded by professional developers, built on top of these foundational component services.
A major enabler for this profoundly powerful shift, particularly for complex enterprises, is a long-awaited industry standard that has been needed since the dawn of the internet: a widely accepted protocol capable of describing business services in a way that can be read and understood by computers, not just people. This critical architectural challenge is now being addressed by initiatives such as The Model Context Protocol (MCP), proposed by Anthropic approximately a year ago. The MCP Protocol has rapidly gained traction because it offers a standardized, machine-readable framework for describing the capabilities and interfaces of software services, allowing AI agents and other automated systems to intelligently interact with them.
If this trajectory holds, the future of the "build vs. buy" dilemma will transcend a binary choice, evolving into a synergistic "yes to both." Companies will continue to procure complex, high-value component services from established SaaS vendors for critical parts of their operations. However, these commercial components will be designed with explicit interfaces and protocols, like MCP, enabling them to be accessed and controlled seamlessly by both human users and sophisticated software entities. Some of this controlling software will manifest as AI agents acting autonomously on behalf of users, while others will be custom, system-integrator-defined workflows generated through Generative AI tools.
These AI agents, which will orchestrate interactions with bought services and other internal systems, are expected to be developed by a diverse set of actors: the SaaS vendors themselves, specialized systems integrators, and increasingly, by end customers leveraging advanced user-programming tools. This represents a profound shift towards a more composable enterprise, where core functionalities are bought as robust services, and differentiating layers of automation and custom workflows are built using AI and user-programming capabilities.
Stakeholder Implications and Broader Impact
The implications of this evolving landscape are far-reaching for various stakeholders:
- For SaaS Vendors: The challenge and opportunity lie in adapting their platforms to be "AI-native" and "AI-consumable." This means not just providing user interfaces for humans, but also robust, well-documented APIs and adhering to protocols like MCP to allow seamless integration with AI agents and custom solutions. Vendors who embrace this composable future will thrive, becoming foundational service providers in an AI-driven ecosystem.
- For IT Departments: The role of IT will transform from being the sole builders and gatekeepers to becoming enablers, architects, and governors of a hybrid software environment. They will be responsible for integrating bought services with user-programmed solutions, ensuring data security, compliance, and architectural coherence. Their expertise will shift towards managing complex ecosystems and providing guardrails for citizen developers.
- For Business Users and Citizen Developers: The "age of user programming" promises unprecedented empowerment, freeing business units from reliance on central IT for many specific, non-core solutions. This will foster innovation and agility at the departmental level. However, this empowerment comes with a crucial learning curve. As non-technical individuals aspire to create solutions beyond simple personal time-savers, they will need to internalize many lessons from the product development world. The most critical lesson is that the technical act of building and delivering a solution is often less challenging than the intellectual endeavor of discovering the right solution to build—one that is viable, valuable, feasible, and usable, and critically, adheres to the thousands of implicit and explicit business rules.
In conclusion, while user programming has largely existed at the margins, the new generation of Generative AI tools is rapidly bringing these capabilities into the mainstream. This is, by and large, a positive development, offering immense benefits to countless individuals and organizations constrained by limited IT resources. The future of business software is not one of either "build" or "buy," but a sophisticated integration of both. It will be characterized by enterprises strategically purchasing complex, robust component services and then dynamically building custom, AI-driven layers on top, orchestrated by intelligent agents and empowered citizen developers. This hybrid model, underpinned by open protocols like MCP, promises a future of unprecedented agility, customization, and innovation, redefining how organizations create and consume technology.
