Making the Case for Data Spaces to Solve AI's Context Challenges in Isolated Systems

Enterprise data management is showing its limits. Modern stacks can store huge volumes, yet the most critical data still sits across clouds, partner systems, and aging tools. AI works with only part of the picture while teams face rising complexity. Centralization was meant to bring clarity, but many organizations now have lakes that function more like holding areas than strategic assets, with data losing its context the moment it leaves its source.

This is the architectural gap that Jayadev Vasantham, Chief Technology Officer at industrial software platform IndustryApps, is focused on fixing. With two decades of experience leading high-stakes, multi-million-dollar digital transformations in the tech and financial services sectors, Vasantham has spent his career navigating the front lines of enterprise data. His point is that the industry's long-standing focus on centralization is becoming unworkable. Instead, he says, the future lies in an alternative model focused on connecting data where it lives rather than attempting to collect it all in one place.

It's not a storage issue but an architectural limit that emerged as enterprises stretched across clouds, partners, and distributed systems. The legacy approach assumed critical information could be gathered in one location and analyzed later, but the reality is that the meaning of data lives in the relationships between systems, not in a single repository that tries to absorb them all.

"The data is scattered and you cannot bring it into one place anymore. Modern enterprises run on collaboration, not isolated stacks, and the old idea of centralizing everything simply no longer fits the way organizations work," Vasantham says. A data space recognizes that distributed data is the norm and builds a unified environment that preserves context, enforces semantics, and connects live information where it resides. It replaces the brittle promise of centralization with a model designed for the way modern enterprises actually operate, giving AI the coherent view it has never had.

• Missing meaning: The limits of the old model become clearest in the AI era. Enterprises may succeed in centralizing data, but they lose the meaning that makes the information usable. As Vasantham puts it, "You might have a table called 'transact two' and have no idea what it actually stores, because each system names and structures its data differently and none of that context survives once you move it." Without that shared understanding, "you cannot run meaningful AI because the system does not know what the data represents," he explains.

• Context is everything: It's a problem baked into earlier architectures, built to collect and store information rather than to explain it. "Those systems were optimized for storage, not intelligence, and they never captured the semantics or relationships that modern AI depends on," Vasantham says. A data space reframes the problem by restoring context and meaning at the source, giving organizations a foundation where AI can operate on data it finally understands.

Vasantham’s vision of a seamlessly connected, queryable ecosystem hinges on a question: How can an organization securely access data residing in a partner's live system? He addresses the common concern by pointing to modern, token-based approaches that create an "invisible AI" layer to securely communicate with siloed systems.

• Whose data is it anyway?: "The biggest challenge people see in federation is authentication, because the data is sitting on someone else’s server," Vasantham says. "But with federated logins and single sign-on, all of this is possible. These systems use tokens so you don't create a direct session on the partner’s server, and that solves the access control problem."

A key part of the appeal is how little disruption the model requires. Instead of replacing systems or rebuilding architectures, a data space layers onto tools most enterprises already use.

• Work with what you've got: "You don't need a huge tech team to build a data space," Vasantham explains. "You can start with what you already have, because the core pieces are already in most environments. You already run enterprise authentication and you already use lightweight query engines, so you are simply connecting what exists rather than building something new." That foundation turns the shift from a major overhaul into a practical extension of current infrastructure.

That blueprint, leveraging concepts like "zero copy" seen in open-source projects, clarifies what comes next. A data space shifts the enterprise from hoarding information to understanding it, connecting datasets where they live and preserving the context that AI needs to operate with confidence. It marks the transition from static lakes to a dynamic, queryable fabric that reflects the way modern organizations actually function. "It's like looking at the sky and trying to find Orion," Vasantham concludes. "You need to connect the dots. A data space can connect the dots for the enterprise world."