Vijay Raina has spent a career at the intersection of enterprise architecture and software design, helping organizations navigate the labyrinth of digital transformation. As a specialist in SaaS technology, he has witnessed the pendulum swing from bespoke legacy systems to the massive wave of “buy-not-build” standardization. However, with the rise of agentic AI and a new era of engineering, Vijay argues that we are entering a post-binary world where the choice is no longer a simple fork in the road between buying a package or building from scratch. In this discussion, we explore how organizations are moving beyond standard vendor roadmaps to reclaim differentiation through custom overlays, re-platforming, and greenfield AI initiatives, all while maintaining the core stability of their SaaS systems. By examining the shifting economics of agentic engineering and the emergence of four distinct technology pathways, Vijay provides a blueprint for a more deliberate, domain-level approach to making confident, forward-looking technology decisions.
How is the emergence of agentic engineering fundamentally altering the traditional economic argument that previously favored packaged SaaS over custom development?
The old math of enterprise technology was rooted in the high “tax” of maintenance; if you built a solution, you owned the burden of keeping it alive forever, which often led to technical debt and eventual stagnation. Agentic engineering is effectively slashing those maintenance costs by automating the heavy lifting of code sustainment and adaptation, making custom solutions feel less like a permanent liability and more like a high-speed asset. We are seeing organizations achieve greater speed and flexibility because these agentic capabilities reduce the manual effort previously required to develop, test, and sustain bespoke logic over long periods. This shift allows a company to move fast without the traditional added risk, turning what used to be a multi-year development nightmare into a streamlined process that mirrors the agility of a digital native. It is a visceral change for IT leaders who are used to the “buy” side of the binary being the only safe bet for reliability, scale, and long-term vendor support.
Given the complexity of modern technology stacks, what are the distinct pathways organizations are now taking to balance vendor reliability with the need for bespoke innovation?
We have identified four specific pathways that allow for a more surgical approach to technology rather than a blunt, enterprise-wide mandate that often fails to meet specific department needs. The first is Vendor-led evolution, which is perfect for non-differentiating, regulated, or standardized capabilities where you want to lean on the vendor’s scale and embedded AI for predictability and compliance. Then there is the Agentic overlay, where you keep the SaaS system as the reliable system of record but layer on a bespoke agentic “brain” to drive unique user experiences and orchestration without disrupting the core foundation. For those feeling suffocated by legacy constraints, the Re-platform pathway allows for replacing high-cost, low-flexibility systems with AI-native solutions that offer total ownership and the potential to optimize cost as usage scales. Finally, there is Greenfield AI, where you build entirely new, unconstrained applications from the ground up to create value propositions and operating models that simply could not exist within the confines of legacy software. Each of these pathways requires a different level of risk appetite and organizational discipline, but together they expand the range of options far beyond the traditional build-versus-buy binary.
When an executive is standing at the crossroads of these four pathways, what specific questions should they be making to determine if a capability deserves a custom build versus a standard vendor solution?
The decision-making process has moved from a generic enterprise level down to the domain level, where we must ask if a specific capability is a true source of competitive advantage or if it is better off being standardized for the sake of auditability. You have to look closely at the “Vendor trajectory”—asking if your current partner is actually a credible innovator whose roadmap aligns with your long-term needs, or if they are evolving too slowly for your specific market. We also look at “Agentic leverage,” questioning how agents can meaningfully enhance or even replace the current capability in a way that provides a significant return on differentiation and control. It is also vital to assess organizational capability; you must determine if you have the governance and FinOps discipline to manage the usage, cost, and value realization of a custom build before you commit. Finally, you have to weigh technical constraints and regulatory exposure to ensure that any move toward flexibility does not unintentionally compromise your requirements for assurance, traceability, and long-term operational simplicity.
As enterprise environments become more distributed with work spanning both vendor and custom systems, what are the primary challenges in maintaining a cohesive data foundation?
The reality today is that work is no longer contained within a single “walled garden” of one vendor; it is spread across a complex web of SaaS platforms, agentic layers, and custom-built applications. This distribution increases the importance of clear accountabilities and a rock-solid data foundation, as any break in the flow of information between a vendor system and an agentic overlay can lead to logic errors and missed opportunities. We are seeing a move toward cross-functional governance where technology, data, and business teams must sit at the same table to manage these interconnected choices and ensure seamless integration across the entire landscape. You cannot just design for today’s reality; you have to design architectures that connect SaaS platforms and agentic layers while allowing for continuous optimization as technologies and vendor capabilities evolve. Without this architectural discipline and strong data foundations, you end up with a fragmented mess that adds more complexity than it solves, regardless of how innovative the individual components might be.
For organizations choosing to re-platform or go greenfield, how does the risk profile change for executive leadership compared to staying with a traditional SaaS vendor?
Choosing to re-platform is a bold move that carries a much higher level of executive accountability because you are essentially betting on your own ability to design, deliver, and sustain a solution better than the broader market can. It requires an early and deep commitment to architectural decisions, as you are moving away from the “safety” of a vendor’s established operating model and into a space where models are less mature and standardized. While this offers the potential to optimize cost and flexibility over time, the “Greenfield AI” approach specifically demands a high risk appetite and the organizational discipline to manage entirely new value propositions unconstrained by legacy logic. Leaders must be prepared to own the long-term sustainment of these systems, ensuring that the governance is in place to handle the evolution of the software as business needs and AI capabilities shift. Success in these pathways is not just about having the best code; it is about having the organizational will to operate within a more flexible, but also more demanding, technological framework that requires continuous ownership.
What is your forecast for SaaS?
I believe we are moving toward a “liquid” enterprise environment where the binary boundaries between “bought” software and “built” intelligence will almost entirely dissolve. In the next few years, SaaS will not disappear, but it will be relegated to being the high-utility “plumbing” of the organization, providing scale and reliability, while the actual competitive soul and differentiation of a company will live in the custom agentic layers they build on top. We will see a massive shift where organizations no longer ask if a platform is “good enough” for their needs, but rather how easily that platform can be controlled, extended, and orchestrated by their own proprietary AI agents. The ultimate winners will be the organizations that master the orchestration of these distributed systems, turning their technology stack into a flexible, living organism that can adapt to market changes in real-time. This is not just a technical upgrade; it is a fundamental reimagining of how enterprise software serves the business, moving from a rigid system of record to a dynamic system of intelligence that balances speed, control, and long-term value.
