The enterprise technology landscape is currently undergoing a structural correction fifty years in the making, as organizations transition from an application-centric model to one defined by data primacy. At the recent Everpure Accelerate event, CEO Charlie Giancarlo outlined a strategic pivot that places data at the center of the enterprise, effectively pushing applications downstream. This vision, while ambitious, reflects a growing necessity driven by the rapid integration of artificial intelligence (AI) and the increasing complexity of global data estates. According to Giancarlo, the traditional architecture—where data is siloed within specific applications—is no longer tenable in an era where AI agents require cross-functional, real-time access to information to provide accurate and actionable insights.
The shift toward data primacy is not merely an architectural preference but a response to the "data sprawl" that has plagued the modern enterprise. For decades, the industry standard has been to build or buy applications—such as ERP, CRM, or HRIS systems—that act as the primary owners of the data they generate. This has resulted in a fragmented ecosystem where "truth" is distributed across hundreds of disconnected platforms. Everpure’s leadership argues that for AI to be effective, this hierarchy must be inverted. Data must become the primary layer, with applications serving as transient tools that interact with a centralized, governed, and semantically understood data core.
The Strategic Context: Beyond the "Boil the Ocean" Approach
While the theoretical "inversion" of the enterprise suggests a radical overhaul, Everpure executives emphasize that the practical implementation is a multi-year journey rather than a sudden replacement of existing stacks. Fred Lherault, Everpure’s Field CTO for EMEA and Emerging Markets, clarified that the company is distancing itself from the "centralization" pitches common among legacy platform vendors. Unlike previous models that required customers to move all their data into a single proprietary repository, Everpure’s approach focuses on federated visibility.
The objective is to create a unified environment where the organization understands where its data resides without necessarily undergoing the massive risk and expense of wholesale migration. Industry data suggests that large-scale data migrations can cost upwards of five times the initial storage budget and take years to complete, often with high failure rates. By advocating for a "map the data, don’t move it" philosophy, Everpure aims to lower the barrier to entry for CIOs who are wary of disrupting mission-critical systems of record.
Rob Lee, Chief Technology Officer at Everpure, noted that this journey is incremental. The immediate priority for technology leaders is to stitch together existing silos and add value through better discovery and governance. This "federated" approach allows enterprises to maintain their current investments while slowly building the foundation for a more data-centric future.
Evolution from Master Data Management to AI-Driven Automation
A recurring question among industry analysts is how data primacy differs from previous movements like Master Data Management (MDM), data lakes, or data meshes. Historically, these efforts have often failed to achieve a "single source of truth" because they relied heavily on manual human intervention. Analysts point out that manual schema mapping is notoriously error-prone and slow, often becoming obsolete by the time the project is finished.
Everpure’s Rob Lee identified two catalysts that distinguish the current era from the MDM cycles of the past. The first is AI-driven automation. Modern machine learning techniques allow for the automated analysis of data layouts and meanings, removing the human bottleneck. The second is the shift from analytical to transactional utility. While MDM and data lakes were designed for "after-the-fact" reporting and analytics, data primacy is designed for "live" transactional flows. AI agents must act within the active business process, requiring a level of data readiness that legacy systems cannot provide.
Furthermore, the proliferation of AI-generated applications is expected to exacerbate data fragmentation. As business units gain the ability to generate niche applications using low-code or AI tools, the number of silos could increase tenfold over the next few years. In this scenario, data primacy stops being a choice and becomes a survival mechanism for maintaining organizational control.
Governance and Liability: The Primary Drivers of Budget
While the long-term vision of data primacy is compelling, the immediate driver for enterprise investment is liability. Prakash Darji, General Manager of Everpure’s Digital Experience business unit, observed that security and governance concerns are the most effective catalysts for unlocking IT budgets in the current economic climate.
The concept of "willful neglect" has become a significant motivator for C-suite executives. During discovery exercises or proofs of concept (POCs), organizations frequently discover exposed or improperly handled data—such as sensitive customer information copied to personal cloud drives or unencrypted spreadsheets. Once these vulnerabilities are identified, inaction creates a legal and financial liability. This realization often leads to the "manufacturing" of budgets to address data classification and security immediately.
Everpure’s strategy involves leading with a data classification offering tailored for governance and security. By tying this to the company’s existing snapshot and recovery capabilities, they provide an automated path to restoration from a "last known good copy" in the event of a breach or corruption. This aligns with broader market trends: a 2023 survey of CIOs indicated that "data trust and governance" are the top two hurdles to AI adoption.
The Multi-Persona Buying Cycle
Implementing data primacy requires significant political will, as it crosses the traditional boundaries of departmental influence. Historically, applications were purchased by departmental leads—such as the Head of HR or the VP of Sales—with IT providing support. Data primacy, however, is a cross-functional middleware project.
Prakash Darji outlined three distinct buying centers that Everpure is targeting:
- The CISO (Chief Information Security Officer): Interested in data classification and the prevention of "willful neglect" through improved security posture.
- The CDO (Chief Data Officer): Focused on agentic middleware and ensuring that data is usable and accurate for AI initiatives.
- The CIO (Chief Information Officer): Concerned with the underlying infrastructure and the long-term architectural efficiency of the estate.
To bridge the gap between being a "storage company" and a "software and governance company," Everpure has shifted its hiring strategy, bringing on sales professionals with experience in software-as-a-service (SaaS) and data governance. This allows the company to engage in high-level conversations with non-infrastructure personas, leveraging its existing multi-million dollar relationships with large enterprises to broaden the scope of its influence.
The Semantic Knowledge Graph: The Intellectual Core
At the heart of the data primacy thesis is the semantic knowledge graph. This layer is designed to describe what data is and what it means, independent of the application that created it. Currently, most data meaning is "locked" within the cryptic code of specific applications. If a different system attempts to access that data, it often lacks the necessary context to use it effectively.
The semantic graph provides an explicit representation of data context. However, executives acknowledge that market education is still in its early stages. Many buyers struggle to define a semantic graph or understand its relevance to AI performance. Prakash Darji noted that common AI issues—such as high costs, low accuracy, and hallucinations—are often symptoms of poor data relevance and a lack of understanding of the inputs. By vectorizing data with proper context, organizations can reduce the "noise" that leads to AI errors, ensuring that agents are retrieving only the most relevant information.
Contextual Governance in the Age of AI Agents
The rise of AI agents introduces new complexities to data access and security. Traditional security models are based on "who" is asking for information. In an agentic world, security must also consider "what" is being asked and the context of the task.
For example, an AI HR agent might be asked to provide a salary recommendation. While the agent may have access to the entire payroll database, it must exercise "judgment" to provide a summary or recommendation without leaking individual records. Encoding this type of contextual governance is significantly more difficult than setting simple read/write permissions.
Fred Lherault highlighted the risks of "context-free" AI actions, citing instances where AI agents have deleted production databases because they lacked the context of a "code freeze" or the destructive nature of a command. As accuracy becomes a production-grade requirement, the ability to explain and reproduce an agent’s actions becomes essential, particularly in regulated industries like finance and healthcare.
The Future Role of Applications and Organizations
If data primacy succeeds, the role of application vendors will undergo a fundamental shift. Rather than being the "holders" of data, applications will be valued for their ability to encode complex business processes. For instance, the value of an HR platform like Workday would shift from storing employee records to its expertise in calculating international tax withholdings or managing complex hiring workflows.
This shift will likely result in a divergence in the software market:
- Process-Centric Applications: High-value tools that manage workflows requiring 100% accuracy (e.g., financial systems).
- Data-Centric Agents: Flexible tools where 99.9% accuracy is sufficient for general productivity and insights.
Organizationally, the division of labor is also expected to evolve. The rise of the "citizen developer"—individuals who can generate agents or apps without a technical background—means that senior engineering roles will shift from writing code to "pruning and cultivating" the semantic knowledge graph. The data function may move toward a "DevSecOps" style of integrated management, though the final structure of the modern data-centric org chart remains to be seen.
Conclusion: Building the Wall from the Bottom Up
The vision of a 50-year architectural inversion serves as a North Star for Everpure, but the path forward is grounded in pragmatic building blocks. The strategy involves helping buyers map their data where it sits, solving immediate governance and liability problems, and gradually expanding toward a centralized semantic layer.
By acquiring companies like 1touch and partnering with established players like IBM, Everpure is positioning itself as a guide for enterprises navigating the "agentic AI" era. The advice from the company’s leadership is clear: start with the most critical data and the most frequent questions, building the foundation of the "wall" before attempting to solve the most complex use cases. In the high-stakes environment of enterprise AI, this bottom-up approach may be the only way to achieve the accuracy and security required for long-term success.
