Salesforce reveals how $8 billion Informatica deal solves enterprise AI's context problem

Salesforce defines Informatica's role in unified data strategy - SiliconANGLE

Salesforce defines Informatica's role in unified data strategy Salesforce Inc. today released details about its thinking in acquiring Informatica Inc. despite that company's apparent overlap with Salesforce's MuleSoft integration subsidiary. Executives said the deal is central to building a unified data foundation for agentic artificial intelligence and resolving long-standing issues of fragmented enterprise context. The acquisition addresses a core obstacle in enterprise AI adoption: lack of business context, Rahul Auradkar, executive vice president and general manager of unified data services for Data 360 and AI Foundations at Salesforce, said in a press briefing. "The models are incredibly intelligent, but they tend to be corporate stupid," he said. "They know almost everything about the world, but very little about your businesses. Without shared enterprise understanding, the AI agents are forced to guess." Informatica Chief Product Officer Krish Vitaldevara said the biggest reason AI projects fail is context gaps caused by data fragmentation, inadequate data or poor data quality. He defined context as "all the background information, relationships and rules that transform raw, isolated data into meaningful, trusted and actionable intelligence." Salesforce executives said Informatica provides that missing enterprise-wide context, while MuleSoft continues to supply the real-time operational signals needed to trigger actions. The combined stack aligns metadata-rich context from Informatica with MuleSoft's event-driven connections, Auradkar said. Informatica's role is to create semantic memory across an enterprise. "Informatica brings the industry-wide, industry-standard, highly regarded [master data management]," he said. MuleSoft, by contrast, supplies "the real-time operational signals such as inventory changes, shipment delays and order exceptions that enable AI to trigger appropriate actions." Data 360, Salesforce's unified data foundation that brings together its Data Cloud, Tableau, MuleSoft and Informatica technologies under one governed architecture, is the unifying layer. "Data360 accesses the activation layer, harmonizing financial operational data directly into Salesforce's metadata using zero copy," Auradkar said. The goal is to provide "a consistent shared understanding of data about products, assets, customers, interactions, locations, suppliers and more. We replace guessing with reasoning." Vitaldevara said Informatica operates at a deeper layer of enterprise-wide data management than MuleSoft, as well as across clouds and on-premises. "We are very much vendor-neutral in the sense that we work with every enterprise application," he said. He noted that Informatica's Cloud-scale AI-powered Real-time Engine, a unified metadata intelligence and AI system that powers its Intelligent Data Management Cloud, processes metadata at massive scale, handling more than 140 trillion customer transactions per month. Salesforce pointed to practical benefits it has seen internally. Auradkar said the company achieved "a 98% reduction in tax adjustments and 20% fewer duplicate accounts," by applying Informatica MDM with Data360 and MuleSoft to its own fragmented account and product data. Executives said customers will soon see Informatica metadata and MDM records flow directly into Salesforce systems without custom code. "That will now not require our customers to work with two separate companies," Auradkar said. Both companies will also align around shared governance and AI-safe access controls. Auradkar said the combined platform strengthens data protections through "masking, attribute-based access control, data lineage and the ability to do data verification." The goal, he said, is to ensure that as AI agents operate across increasingly heterogeneous environments, Salesforce can "respect the governance and the access of data that exists in other systems." Vitaldevara said Informatica's mission remains unchanged. "We have always said we are the Switzerland of data management, and now we are going to be the Switzerland of data and the Switzerland of Ain," he said.

Two become one - Salesforce and Informatica map their future together (with MuleSoft as part of their $8 bn data thruple)

We all know data alone is not enough, and context is a new currency in the world of agentic AI. A simple assertion from Krish Vitaldevara, Chief Product Officer at Informatica but one that might sum up why Salesforce has been happy to shell out $8 billion to make the firm the latest outpost of its agentic empire. It was back in May that the takeover of Informatica by Salesforce was confirmed, a purchase that was successful on its 'second time lucky' attempt after an earlier putative bid fell through on price. Although Salesforce CEO Marc Benioff, never knowingly under-selling anything, pitched it at the time as a deal that had been two decades in the waiting: We really love the company, we love the people, we've loved the leadership. And I have to say, we've probably spent the last 20 years discussing how to bring the companies together. Flash forward six months and with the all regulatory I's dotted and T's crossed, the deal was closed in mid-November. Fortunately, given the price tag, Benioff has lost none of his enthusiasm, declaring on the recent Salesforce earnings call: Informatica with Data 360 [formerly known as Data Cloud], MuleSoft, I mean, that is taking everything to this new level...Our data infrastructure is incredible - we call it our data foundation. It [is composed of] three key things: Informatica, Data 360, and also MuleSoft. Together I think it will do about $10 billion next year in business. OK, that's the big picture, but now the deal has closed, the devil is in the detail as to how Informatica will be integrated - appropriately enough - into Salesforce in practice. Vitaldevara starts with a high-level pitch: It's important to ground ourselves in the core problem every enterprise is trying to solve today, how to be successful in their AI and agentic journeys. To understand this, though, we need to understand why most of these AI projects fail. We all know it's not the model, it's the data. Almost 80% of AI projects fail, and they cite data fragmentation, inadequate data or poor data quality as the key reasons. It underscores the importance of quality, consistency, and meaning of the data underneath the model to be able to get full value. We also are very aware that different systems store different fragments of meaning, each with its own language, its own rules and its own truth. AI cannot see the full picture because context is scattered, stale or inconsistent. AI models are intelligent, but they almost know nothing about your businesses and need significant domain specific knowledge. The biggest bottleneck that we see for businesses becoming agentic enterprises is this lack of trusted context. What all this means, he contends, is that agentic enterprise context is all the background and relationships and rules that transform raw, isolated data into meaningful, trusted and actionable intelligence for an AI system,: It is the metadata. It's the lineage, relationships, governance. These are the things that tell AI what a product is, how a process works, where the data comes from, and whether it can be trusted. Context is the digital equivalent of AI's working memory and situational awareness? It is the institutional knowledge. The magic really happens when we use products like Informatica Data Management Cloud to turn raw data into trusted context. And it is this trusted context that enables accurate AI, replacing guessing or even rule-based models, with reasoning-based models. And this is where Informatica really is. Picking up the story from the Salesforce perspective, Rahul Auradkar, EVP & GM, Unified Data Services, Data 360 & AI Foundations, bluntly argues that while today's AI models are "incredibly intelligent", in practice they are "corporate stupid": It knows almost about everything about the world, but very little about your business. Without the shared understanding of the enterprise, AI agents are forced to guess. What AI agents see is fragments, whether it's a batch in Manufacturing, an order in Commerce, or a ticket in Support, but they lack trusted enterprise context to connect these pieces in a meaningful way. That's where context is the new currency. He cites a personal example of the problematic consequences of data fragmentation with an anecdote about researching and buying a car with his daughter a few months ago, her first such purchase. The research journey led to a purchase and that journey should have been contextualized in the systems, he explains, but in fact it was not mapped. This had negative consequences: What happened here was the systems capture the research, and another system captures the purchase data, but they treat them as independent events. Now, marketing automations and marketing agents promptly try to sell me and my daughter the same car again within a month. This leads to a significant amount of millions of dollars wasted, if you may, in marketing spend. The data in this case is accurate. It's available, it's high quality, but in fragments. The context was wrong, so it could not be used for additional business engagement. That sort of thing won't happen using the triumvirate of Data 360, MuleSoft and Informatica, he suggests: We are delivering the most sophisticated data foundation to power the entire AI trusted enterprise context, We give every part of the enterprise a consistent, shared understanding of its data, data about products, about assets, about customers, about interactions, locations, suppliers and more. And when every system, workflow and agent operates from the same context, decisions become faster, AI becomes more accurate, and automation becomes more reliable For his part, Vitaldevara checks in at Holiday Inn Vacations to demonstrate how all this can work better: They have been focused on this multi-year transformation to modernize operations and enhance member experience. As part of this digital transformation, they upgraded many core customer-centric systems, including Salesforce and they unified customer data across the enterprise by using Informatica. The challenge they faced is pretty typical of most enterprises who are doing these sort of transformations. The data is scattered across several systems, each of the systems was critical for managing operations across the customer journey, and they were largely siloed, which makes it incredibly difficult to get the common view of customers across the stack. Fragmented data also translated into more data quality issues Put it all together and it was an incredibly hard journey for them to deliver real-time, personalized experiences that customers increasingly expect. But by unifying all of this data with Informatica and Salesforce, they could rapidly implement that 360 degree view of its 350,000 plus members across their portfolio, and they were able to drive greater personalization across online and offline touch points. With the combination of with a combination of Informatica and Salesforce, now we have the industry's best data catalog by combining the metadata model that we had, which made us the number one AI CRM, with that of informatica's enterprise wide catalog. Twenty years in the making - the future starts today.