Enterprise AI’s next big challenge is getting AI systems to understand the businesses they are supposed to run. A large language model (LLM) can summarize a customer complaint or recommend what a salesperson should do next. An AI agent can go further, taking action within an enterprise system without waiting for an employee to approve each step. However, it needs to understand what products the customer owns, whether there is an unresolved service problem, and if the company’s policies allow it to do.
That information usually exists, but it is often scattered across the enterprise.
Reltio, an enterprise data management company, describes this approach as “context intelligence”. CEO Manish Sood describes the resulting architecture as a system of context: a layer that connects information across an enterprise so AI systems can understand how each piece of information relates to one another, and what decisions the business can make with that context.
“Agentic AI needs access to the data that sits in all the different existing IT systems across an enterprise. But understanding that information requires more than knowing where it is. It requires the actual data inside those systems. That combined capability, both data and metadata together, is the notion of context intelligence,” he told me. “How a company like CarMax runs its business is very different from how a life sciences company like Eli Lilly runs its business, and that context resides in their existing IT landscape. Context intelligence acts as the bridge between that landscape and the new capabilities AI is creating.”
SAP now owns that layer. The enterprise software giant completed its acquisition of Reltio in May, saying the deal will strengthen Business Data Cloud and make it a more interoperable enterprise data platform capable of connecting SAP and non-SAP data for analytics and AI.
“Where the meeting of the minds happened with the SAP leadership team was around this notion of the autonomous enterprise and the vision that they laid out. It’s no longer about being an ERP company. It’s about how we are going to define the future of the enterprise,” Sood says. “The notion that we could bring Reltio as a capability to the customer base SAP has (more than 400,000), in a much smaller time frame and at a very fast and accelerated speed, was a compelling proposition.”
Before Reltio, Sood helped build Siperian, an enterprise master data management company that Informatica acquired in 2010. He founded Reltio the following year to tackle the same underlying problem from a cloud-first approach, moving beyond traditional MDM toward a more connected view of enterprise data. SAP’s acquisition now gives Sood a second experience of watching a company he built become part of a much larger enterprise software platform.
But the acquisition also creates a new tension. If context becomes a critical layer in enterprise AI, can Reltio remain neutral while owned by one of the world’s largest enterprise software companies? Sood argues that SAP understands why Reltio needs to remain open.
“SAP realized that every customer has a heterogeneous environment, and that is why this kind of capability is essential,” he says. “Customers are also asking for that capability to remain heterogeneous, and that is why the whole notion of keeping it open is important.” He also notes that AI itself makes a single-vendor approach impractical. “At the end of the day, AI requires a multi-channel, multi-application type of data set, because one of the key benefits AI can deliver is cause-and-effect correlation.”
Enterprise AI Needs Context at Operational Speed
Sood believes that the context layer cannot simply be an analytical repository. It needs to deliver trusted information quickly enough to influence an operational decision as it happens, and AI agents intensify that pressure because they can make decisions continuously rather than waiting for employees to review reports.
He points to Eli Lilly, Novartis and AstraZeneca as companies rethinking their operations across the drug development and commercialization lifecycle, from R&D and clinical trials to bringing new medicines to market and engaging directly with patients. Sood says that pace of change is accelerating, forcing companies to rethink how they operate. AI is increasingly part of that transformation, but he says that the model is only one component.
“A lot of the investments they are making are going to be AI investments, all the way from drug discovery to how they want to engage with patients to create personalized treatment outcomes and benefits,” he says. “That is where they are putting in Reltio-type product capabilities to create that system of context, or the context intelligence layer, that will start to inform some of these new methods they are deploying inside their organizations.”
Sood says Reltio was designed from the outset for real-time operational use, with the expectation that AI will only increase the need for faster access to trusted data. He sees this as a key difference between Reltio and analytical platforms such as Snowflake and Databricks. Reltio is designed to give AI access to the freshest, most trusted enterprise data at the point of engagement, where decisions may need to happen in milliseconds.
“Most of the analytical platforms, whether it is Snowflake or Databricks, come from the analytical side of the house,” Sood tells me. “They are not prepared or oriented for this millisecond real-time operational speed and scale, and that is why they are still analytical platforms. They are not the foundation on which you build applications.”
His distinction is partly about where enterprise technology starts. “Applications are where SAP’s expertise is,” he says. “SAP has developed and delivered capabilities that you run your business on. Analytics is always a sidecar. Nobody in the enterprise starts with, ‘I don’t have any applications. Let me put together a lakehouse first, and then I will build my applications.” Sood believes AI will pull operational applications and analytical platforms closer together. As agents take on more business processes, he argues, the context layer needs to sit close to operational systems and deliver trusted information in real time.
That puts Reltio in a growing race with companies approaching the context problem from different parts of the stack. Palantir has made its ontology central to its enterprise AI strategy, while Snowflake and Databricks are pushing beyond analytics into operational AI. Major application vendors are also building their own semantic, data and agent layers. Sood sees the growing competition as “validation of the vision that we had from the very beginning and founding of Reltio that there was a need for a system of context.”
But he acknowledges that a growing market does not guarantee Reltio will win. The company still has to prove what sets its approach apart, with operational speed and implementation at the center of its differentiation.
The Control Layer Verdict
Sood bets that enterprise AI will depend less on smarter models than on the context that lets them make reliable decisions. SAP gives Reltio the scale to make that context layer a bigger part of enterprise AI, but it also creates the central tension at the heart of the deal. Can Reltio become the connective layer across an enterprise without becoming another way for SAP to pull customers deeper into its ecosystem? If context becomes as important as the models making use of it, control over that layer will matter. The technology may prove that AI needs context; The next few years will show whether customers can trust the company that controls it.
