Vijay Raina brings a seasoned architectural perspective to the shifting landscape of enterprise technology, specializing in the structural design of SaaS and software tools. As the industry grapples with the fallout of traditional subscription models, Raina provides essential clarity on how AI is not just an add-on, but a fundamental replacement for the software frameworks we have relied on for decades. In this discussion, we explore the transition from horizontal platforms to vertical, AI-native solutions that prioritize outcomes over seat counts.
Our conversation centers on the significant market correction that saw a $300 billion single-session wipeout, signaling the end of the traditional SaaS peak. Raina elaborates on the “death” of per-seat pricing, the rise of “headless” models where AI agents act as the primary users, and the strategic importance of the “Three Ds”—Distribution, Domain expertise, and proprietary Data. We also delve into the “Human-in-the-Loop” philosophy and how the industry is moving from competing for IT budgets to capturing a much larger slice of the $2 trillion white-collar labor market.
The enterprise software market recently saw a massive valuation wipeout, signaling a move away from traditional per-seat models. How is the emergence of “headless” AI models contributing to this loss of predictability that investors once cherished?
For a long time, investing in SaaS was the closest thing to a sure bet because the returns were so predictable. You would build a horizontal, cloud-based platform, target a massive market, and grow your revenue simply by expanding the number of seats or users. However, the $300 billion single-session wipeout we witnessed in January serves as a stark leading indicator that this era has reached its peak. The reason for this sudden instability is the rise of AI agents—what we call “headless” models—where the software itself becomes the user rather than a human employee sitting at a desk. When an AI agent can autonomously navigate a workflow, the traditional metric of counting “seats” becomes irrelevant, causing investors to retrench and reconsider the entire valuation framework of the industry. This shift is fundamentally upending the scalability that once made names like Salesforce and Workday the darlings of the public markets.
If the traditional per-seat pricing model is indeed “evaporating,” what specific metrics or pricing structures should technology companies adopt to survive this transition?
As AI agents begin to generate the bulk of software usage, the old way of charging per license is no longer sustainable. We are seeing a move toward models that charge for the actual work performed or the specific outcomes achieved, often referred to as ROI-based pricing. For example, a legal AI platform shouldn’t just charge for a login; it should charge for every contract it successfully drafts, essentially capturing a fraction of the labor cost it is replacing. We see similar shifts in spend management where software takes a percentage of the overages it identifies, or chargeback applications that take a fee based on the value of the funds they recover. A company that once required 100 CRM licenses might soon only need 50 because AI is handling the bulk of the manual entry and follow-up, forcing vendors to find value in the results rather than the headcount.
You’ve mentioned that generic horizontal SaaS is becoming a liability. Why are specialized vertical niches suddenly more defensible than the massive, all-in-one platforms of the past?
Horizontal SaaS is increasingly vulnerable because much of its value was built around simple workflows that AI can now handle autonomously without much effort. If your product is essentially a “wrapper” around a basic task like form building, project management, or social media scheduling, your value proposition is being compressed at an alarming rate. In contrast, vertical specialists are protected by what I call the “Three Ds”: Distribution through a long-standing customer base, Domain expertise in complex or regulated industries, and proprietary Data that is inaccessible to generic frontier models. When software is deeply embedded in the specific compliance requirements and terminology of a niche like insurance underwriting or bank loan performance, it creates a moat that is incredibly difficult to cross. Exporting a list of contacts is easy, but you cannot simply export the institutional logic and historical knowledge baked into a specialized vertical model.
When we talk about proprietary data moats, how does the integration of specialized industry workflows create higher switching costs compared to the previous generation of software?
In the previous era, switching costs were often just about the headache of migrating data from one cloud to another. Today, the cost of leaving a vendor is becoming much higher because the software is integrated into the very logic of how a business operates. If a legal contract repository or an insurance platform has spent years embedding specific regulatory constraints and corner cases into its AI models, moving to a different provider means more than just a data transfer; it means rebuilding a complex web of historical knowledge and operational judgment. Customers stay because the cost of retraining their systems and finding a new vendor who truly understands their specific world is prohibitively expensive. These assets create a level of “stickiness” that a generic SaaS contract could never produce, as the software becomes a core component of the company’s decision-making engine.
There is a growing emphasis on “Human-in-the-Loop” (HITL) as a defining model for the next decade. Why is the combination of software and human judgment so critical in high-stakes verticals like healthcare or defense?
While AI is exceptional at handling routine and repetitive tasks, there are certain areas where the cost of an error is simply too high to leave to an unsupervised algorithm. In verticals like healthcare, cybersecurity, construction, and defense, we are seeing a shift where agentic intelligence is paired with human judgment at the most critical points in a workflow. This “solutions-centric” approach changes the fundamental nature of a software company from a mere tool provider to a strategic partner that manages onboarding, design, and quality control. Every time a human interacts with the system to refine a high-stakes decision, it creates a feedback loop that makes the product smarter and deeper. By treating human services as a compounding asset rather than a cost of implementation, these companies accumulate a level of institutional trust that is impossible for a “pure” SaaS product to replicate.
The potential market for AI-native software is described as being much larger than the traditional SaaS market. How does moving from IT budgets to labor and productivity budgets change the conversation for software founders?
The shift we are seeing is a move from competing for a slice of the IT budget to going after the much larger $2 trillion white-collar services market. Traditional SaaS was always limited by how much a company was willing to spend on its technology stack, but AI-native platforms are now competing for the labor, compliance, and risk budgets. When you consider McKinsey’s projection of a $6 trillion annual productivity opportunity from AI transformation, it becomes clear that the pie is significantly larger than anything we saw in the enterprise software era. Winners in this space won’t just be bolting AI onto old products; they will be firms with deep subject matter expertise that use AI to collapse the boundary between software and services entirely. This is a fundamentally different kind of company that creates value with every data asset it accumulates, making it worth considerably more than its predecessors.
What is your forecast for the future of independent, owner-operated software firms in this new AI-driven landscape?
I believe we are entering a golden age for independent, owner-operated businesses that have spent years mastering specific industry domains without the pressure to become generic horizontal giants. These firms are uniquely positioned to leverage their deep historical data and specialized workflows to build AI-native tools that larger, more diluted competitors simply cannot match. While the $300 billion wipeout felt like a catastrophe for the old guard, it actually clears the way for specialized players to capture the $6 trillion productivity opportunity by focusing on outcomes rather than seats. The future belongs to those who stop viewing themselves as software vendors and start seeing themselves as the automated backbone of their respective industries. If you can own the data and the logic of a specific vertical, you won’t just survive the death of SaaS—you will define the next era of enterprise value.
