New Salesforce engine designed to bring trustworthy AI to the enterprise

Salesforce has introduced a major enhancement to its core data platform, revealing a unified data engine that integrates Data 360, Informatica and MuleSoft. The intent is to provide enterprise AI with reliable business context so systems stop hallucinating and start reasoning — a change that could be especially impactful for marketers.

AI hallucinations remain a major obstacle to enterprise adoption. Salesforce says more than 80% of enterprise AI initiatives never progress beyond the demo phase because the models aren’t anchored in the organization’s own data, definitions and workflows. Lacking that grounding, agents misread signals like “customer ID,” “order status” or “return request,” which results in bad insights or wrong actions.

Dig deeper: 5 ways AI changed marketing strategy in just one year

“AI without context is just guessing, or hallucinating,” said Rahul Auradkar, Salesforce’s EVP and GM, Data 360. “By combining Informatica’s metadata with MuleSoft’s real-time signals, we replace guessing with reasoning. We are giving AI the grounding it needs to operate safely — ensuring that when an agent acts, it does so with the full weight of enterprise truth behind it.”

Building context across three layers

Salesforce’s unified data engine rests on three core components:

1. Enterprise understanding through metadata: Data 360 now incorporates Informatica’s master data management (MDM) capabilities for entities such as products, suppliers and assets. This gives AI agents a consistent vocabulary and shared understanding, so when a model encounters “SKU-123,” it recognizes it as the same item as “Part A” in another application. The integration also brings in data lineage, enabling AI to check how current and reliable data is. Combined with a broad data catalog that spans on-premises, cloud and legacy environments, it creates a detailed map of the company’s data ecosystem.

2. Real-time awareness via MuleSoft: MuleSoft supplies continuous operational signals — from shipping delays and inventory shifts to customer behaviors. AI agents can respond to what’s happening in the business right now, instead of relying on outdated snapshots.

3. Unified context with zero-copy architecture: Data 360 serves as the consolidated memory layer, blending Informatica’s historical context with MuleSoft’s live signals. This shared context is made available to AI agents using a zero-copy model, where data is accessed rather than replicated, cutting both storage overhead and latency.

This data foundation underpins Salesforce’s broader Agentforce 360 platform — a four-layer architecture built to power enterprise-ready AI agents. If it performs as described, AI agents managing tasks like refunds, order changes or personalization should be able to operate with the same situational clarity as experienced staff.

What matters to marketers

For marketers, this goes far beyond an IT infrastructure upgrade. Consistent, trustworthy data is what makes AI agents practically useful across campaigns, customer journeys and touchpoints. With real-time visibility, standardized definitions and accurate customer profiles, marketing agents can suggest content, refine targeting and tailor messages without veering off course.

Dig deeper: How to stop AI decisions from repeating human biases

It also supports stronger, more transparent AI governance — increasingly important as regulations tighten and customers expect more insight into how their data is handled.

Salesforce’s new data engine is available now within Data 360, Informatica, MuleSoft and the Agentforce 360 platform. For organizations intent on scaling AI, this could be one of the year’s most consequential technical releases — not because it’s fashionable, but because it tackles enterprise AI’s most stubborn challenge: missing context.

The post New Salesforce engine aims to make enterprise AI trustworthy appeared first on MarTech.

Continue reading →