Salesforce Unveils Trustworthy AI Engine for Enterprises
▼ Summary
– Salesforce has launched a unified data engine that integrates Data 360, Informatica, and MuleSoft to provide trusted context for enterprise AI systems, aiming to reduce AI hallucinations.
– A major barrier to enterprise AI adoption is that over 80% of projects fail because AI lacks grounding in a company’s specific data, definitions, and processes.
– The engine’s three pillars are enterprise understanding via Informatica’s metadata, real-time awareness from MuleSoft’s operational signals, and unified context using a zero-copy architecture to merge and share data.
– This data layer supports Salesforce’s Agentforce 360 platform, enabling AI agents to handle workflows like refunds and personalization with clarity and accuracy.
– For marketers, this unified data is crucial for deploying effective, real-time AI agents in campaigns and customer interactions while also improving AI governance and compliance.
Salesforce has introduced a major upgrade to its core data platform, launching a unified engine that integrates Data 360, Informatica, and MuleSoft. This move aims to provide enterprise artificial intelligence with the reliable, contextual foundation necessary to move beyond generating unreliable outputs and toward making accurate, reasoned decisions. For marketing professionals, this development represents a pivotal step in deploying AI that can be trusted with critical customer interactions and campaign management.
A primary obstacle preventing widespread enterprise AI adoption is the tendency for systems to produce incorrect or fabricated information, often called hallucinations. Salesforce reports that more than 80% of enterprise AI projects stall after demonstrations because the underlying models lack proper grounding in a company’s unique data, definitions, and operational processes. Without this essential context, an AI agent might misinterpret fundamental signals like a “customer ID” or an “order status,” leading to erroneous insights and potentially costly actions.
“When AI lacks proper context, it’s essentially just making guesses, which leads to those hallucinations,” explained Rahul Auradkar, EVP and GM of Data 360 at Salesforce. “Our integration of Informatica’s metadata with MuleSoft’s live signals replaces guesswork with genuine reasoning. We are providing AI the grounding required for safe and effective operation, ensuring every action an agent takes is backed by verified enterprise truth.”
This new unified data engine is constructed upon three foundational layers to build that critical context. First, it establishes enterprise understanding through enhanced metadata. By incorporating Informatica’s master data management into Data 360, the system gives AI agents a shared vocabulary for key entities like products and suppliers. This means an AI model recognizes that “SKU-123” in one system is identical to “Part A” in another. The integration also brings data lineage capabilities, allowing AI to verify data freshness and trustworthiness, supported by a comprehensive catalog spanning cloud, on-premise, and legacy systems.
Second, the engine gains real-time awareness via MuleSoft. This component feeds live operational signals, such as inventory changes, shipment delays, or immediate customer actions, directly into the AI. Consequently, agents can act based on the current state of business operations, not outdated information.
Third, a unified context layer is created using a zero-copy architecture. Data 360 functions as a harmonized memory layer, merging historical context from Informatica with the real-time signals from MuleSoft. This combined context is shared across all AI agents without duplicating the underlying data, which reduces storage costs and system latency.
This robust data layer serves as the base for Salesforce’s broader Agentforce 360 platform, a multi-tier architecture designed for enterprise-grade AI agents. If successful, agents managing workflows like processing refunds, adjusting orders, or personalizing customer experiences could operate with the clarity and accuracy of experienced human employees.
For marketing teams, this advancement extends far beyond back-end IT. Unified and trusted data is the essential ingredient for making AI agents genuinely effective across campaigns, customer journeys, and interactions. With access to clean definitions, verified customer records, and real-time status updates, marketing AI can recommend content, fine-tune targeting, and personalize communications without deviating from accurate information.
Furthermore, this approach establishes a stronger foundation for confident and transparent AI governance. As regulatory compliance demands increase and customers seek greater clarity on data usage, having a verifiable, context-rich system becomes critically important.
The new unified data engine is now available within Data 360, Informatica, MuleSoft, and the Agentforce 360 platform. For organizations committed to implementing AI at scale, this update may be one of the year’s most significant technical developments. Its importance lies not in following a trend, but in directly addressing the most persistent challenge in enterprise AI: the fundamental lack of reliable context.
(Source: MarTech)
