Without good data, AI can’t add much value to organizations. That’s a given. However, it’s not easy to optimize the data pipeline. Everpure aims to offer a solution to this with Data Intelligence and Data Stream. These additions to the platform are also intended to ensure that organizations can finally make a real shift toward a data-centric architecture.
Everpure has been working for quite some time on adding more and more data management capabilities to the platform. Think of features such as the Unified Data Plane and the Intelligent Control Plane that sits on top of it. These components are crucial parts of the Enterprise Data Cloud architecture that the company announced last year during Pure Accelerate. That architecture, in turn, is a logical consequence of what became possible once Everpure had fully developed Pure Fusion after several years.
These developments enabled Everpure to view and manage the underlying storage as a single large storage pool. An additional layer of abstraction on top of all the hardware Everpure delivers within its FlashArray and FlashBlade portfolios meant the company could then do some clever things with it.
Today, many of those smart innovations revolve around AI. More specifically, this involves searching for, finding, and preparing data to deploy AI as successfully as possible. Everpure Data Intelligence and Everpure Data Stream are designed to make this possible.
Everpure Data Intelligence
To a certain extent, Everpure Data Intelligence isn’t really anything new. As far as we can tell, it’s primarily the new name for 1touch.io, the company Everpure acquired earlier this year. With Data Intelligence, it’s possible to locate all the data that organizations have scattered across a vast number of sources. This can include both unstructured and structured data. Furthermore, it doesn’t matter where this data is located. In other words, it isn’t limited to Everpure hardware.
In addition to providing insight into how much data organizations have, Everpure Data Intelligence is also designed to automatically provide insight into the type of data it finds. For example, it can identify and classify sensitive data such as PII. It also immediately shows how this data moves throughout the organization. This should make it possible to thoroughly map out an organization’s data landscape. This type of governance is crucial for compliance.
A third key component of Everpure Data Intelligence is particularly valuable for AI purposes: it provides context. Without context, data cannot be meaningfully utilized to enable AI agents to perform their tasks. Only when the agents understand the significance of the data for actual processes can they get to work. Data Intelligence thus builds, as it were, a semantic knowledge graph that AI agents can use.
Everpure Data Stream
Today’s second announcement isn’t exactly new either. Everpure Data Stream was already announced about six months ago. However, as of today, it is generally available. Data Stream is a service that Everpure has built on the Nvidia AI Data Platform and is designed to ensure that data can be delivered in the optimal format to AI agents and applications.
In other words, Everpure Data Stream is a data preparation addition to Everpure’s platform. It vectorizes the data to make it suitable for use by AI agents and applications. Not only that, but Data Stream does this automatically. According to Everpure, this should reduce the time it takes to properly prepare data from months to minutes. The service also addresses issues such as security and compliance. The streams that are set up can be managed in fine detail. This is intended to ensure that no information leaks outside the company’s own network.
Efficient AI based on the right data
The combination of Everpure Data Intelligence and Everpure Data Stream is certainly intriguing. The former ensures that only the right data is made available for AI purposes, while the latter handles the efficient preparation and vectorization of that data. It also promises that this process occurs continuously. So if anything changes in the underlying data, Data Intelligence picks up on it immediately and makes adjustments where necessary.
The potential of Everpure Data Intelligence and Everpure Data Stream becomes clear when we hear some statistics about data from Fred Lherault, Field CTO EMEA at Everpure. According to him, only 15 percent of all the data organizations hold is actually useful. Of the remaining 85 percent, half is so-called dark data, data that was once stored but is never used. One might also wonder to what extent organizations are even aware that they possess this data.
Cost-savings can be substantial
It goes without saying that identifying the 15 percent of data that is actually useful is extremely important, especially when it comes to making that data available to AI agents and applications. Not only because it yields much better results when only useful data is fed into AI models, but also because cost plays a significant role here. Vectorizing all kinds of useless data and then feeding it into AI models is extremely costly. Additionally, your data pipelines will also have the necessary governance and compliance if you know exactly what’s going into them and do so in a structured manner.
At present, Data Intelligence and Data Stream are still largely standalone additions to the Everpure platform. That is to say, organizations can use both, but Everpure still needs to build some connections between the two to truly turn them into an end-to-end data pipeline, Lherault notes. In addition, we must also assume that Everpure Data Intelligence is actually able to find all useful data. That will also be an ongoing process, because as new sources emerge, new connectors will be needed as well.
From app-centric to data-centric
Everpure Data Intelligence and Everpure Data Stream indicate the direction Everpure is moving in as a company. Rob Lee, Everpure’s CTO, emphasized during a session we attended that the company will certainly continue to develop the underlying hardware. Given the tremendous growth figures of recent years, it wouldn’t be very wise to lose focus on that. We should view the developments in data management and data intelligence primarily as an expansion. It gives Everpure the opportunity to eventually offer the entire data layer to customers, from hardware to software.
More broadly, with today’s new additions to the platform (both available immediately), Everpure also aims to signal something else. We should view Data Intelligence and Data Stream as a sign of the times. That sign is that we need to move away from architectures built around applications and toward data-centric architectures. In other words, if organizations ensure that their data layer and access to it are in order, they can be successful in the markets in which they operate.
The idea is that a data-centric approach frees the data from the applications in which it has been trapped until now. That approach leads to massive data fragmentation, the same data scattered across all sorts of different applications, with associated extra costs and, in general, little clarity about where things are located. In a data-centric model, all data resides in a shared system of record with corresponding governance. The data includes the necessary metadata that clearly indicates what it is. It carries its own meaning, as it were. Applications and agents have access to the data and can add data, but they never own it.
Will it work this time?
The desire to move toward data-centric architectures sounds familiar to us. We’ve been hearing about the need to implement data-centric architectures for nearly a decade now. It still hasn’t happened, no doubt partly due to the power that SaaS platforms wield in the market. Is this just old wine in new bottles, or is the wine actually new?
Given the stories about the “SaaSpocalypse” that have been circulating over the past six months, the timing for a transition from app-centric to data-centric now seems better than it did 5–7 years ago. The stories about the death of SaaS are grossly exaggerated, by the way (podcast in Dutch), but there is a grain of truth to them. That truth lies primarily in the closed nature of the applications. That seems to be slowly but surely becoming a thing of the past.
For example, with Headless 360, Salesforce has taken a clear step toward opening up its system of record for external access. It places less emphasis on its own front end, which, according to Everpure, acts more or less as a prison gate for the data held captive within an app-centric architecture.
Data-centricity is en vogue
If we extrapolate what Salesforce (but also, for example, ServiceNow) is doing in terms of its headless initiatives, we could argue that they also recognize that a data-centric approach is better than a strictly app-centric one (mind you, still all in their own environments, not distributed). In that regard, the stars are now aligned more favorably than before for data-centricity. Provided Everpure can deliver on its promises, the transition to a data-centric architecture might just succeed this time.
On the other hand, if companies like Salesforce also start actively getting involved, the market could become very crowded, and Everpure will have to work hard to further develop this architecture and make it increasingly attractive. Add to this the competition from incumbent data intelligence companies, and Everpure has its work cut out for itself. The fact that it has a complete underlying portfolio of hardware and as such can, on paper, deliver a true full-stack data (intelligence) platform, could come in very handy indeed.