Giving AI a Memory: Salesforce and Informatica Empower Your AI Agents to Reason Accurately and Take Action You Trust

If you’ve ever attempted to move an AI agent from a cool demo to production-ready, you’ve likely encountered the “context gap.” It is one of the top reasons why more than 80% of AI projects fail.

AI models are intelligent, but they are corporate-stupid. They know almost everything about the world but almost nothing about your business. Without a shared understanding of your enterprise, the AI agent is forced to guess. It sees your data as high-entropy noise — a “customer ID” is just a string of numbers; a “shipment status” is just a phrase. It sees fragments — a batch in manufacturing, an order in commerce, a ticket in support — but lacks the trusted context required to connect those pieces into a meaningful story.

Think about it: You wouldn’t expect a new employee to make critical decisions without understanding your business history, rules, and current status. Your AI agents are no different. They need that same shared context to operate effectively, and today we’re changing the game to provide it.

By combining Salesforce and Informatica’s enterprise metadata with Data 360’s harmonized sub-sec real-time context and MuleSoft’s integration strength, we replace guessing with reasoning. - Rahul Auradkar, EVP & GM – Unified Data Services, Data 360 & AI Foundations at Salesforce

By expanding our Data 360 portfolio to include Informatica to improve the quality and clarity of your core business data, and MuleSoft to enrich it with operational signals, we’re moving beyond simple data integration to a unified engine that delivers the trusted context your apps and AI agents need to operate intelligently and safely. 

“AI without context is just guessing or hallucinating,” said Rahul Auradkar, EVP & GM – Unified Data Services, Data 360 & AI Foundations at Salesforce. “By combining Salesforce and Informatica’s enterprise metadata with Data 360’s harmonized sub-sec real-time context and MuleSoft’s integration strength, we replace guessing with reasoning. We aren’t just unlocking trapped data; we are giving AI the grounding it needs with trusted context to operate safely — ensuring that when an agent acts, it does so with the full weight of enterprise truth behind it.”

To make this possible, Data 360 combines three critical components that give Agentforce the context and understanding of a human expert.

1. Enterprise Understanding (Data 360 and Informatica): First, apps and AI agents need to understand the definitions of your business. Informatica feeds Data 360 with the deep metadata intelligence required to define your core business entities and their relationships to each other.

• Enterprise MDM: While Data 360 creates golden records for your customers, Informatica brings Master Data Management (MDM) for the rest of the business objects — products, suppliers, assets, and more. This creates a shared data vocabulary, helping to ensure your apps and AI agents understand that Part A in the warehouse is the same as “SKU-123” in the catalog.
• Data Catalog: We’re combining Salesforce’s metadata model with Informatica’s enterprise-wide data catalog to create a single view of your data across systems. This gives your apps and AI agents a map of assets residing in your on-premise systems, data lakes, and legacy infrastructure.
• Data Lineage: This acts as the data verification layer. It traces the full journey of your data from origin to consumption, showing every transformation along the way. That means an AI agent knows where a piece of data came from, how it was produced, and whether it’s fresh enough to trust.

2. Connected Signals and Action (MuleSoft): Second, apps and AI agents rely on signals from every system to understand context. MuleSoft connects those systems and surfaces the operational signals that let AI interpret what’s happening and trigger secure, meaningful actions.

3. Unified Data Context (Data 360): Finally, we need a place to synthesize and activate this information. Data 360 is the memory. It combines the rich business context from Informatica with the operational signals from MuleSoft into a single, harmonized context interface for your apps and AI agents.

• Using Zero Copy, it makes this rich context available to every agent without moving data. This allows apps and AI agents to access any data in your enterprise without incurring the storage costs associated with copying it.
• This helps to ensure that whether an AI agent is analyzing a trend or executing a product return, it is operating on a single, shared, and up-to-the-second view of the customer, the business, and its policies.

Agentforce 360 is built on four layers, and it all begins with the data layer. This layer gives AI the trusted enterprise context it needs to understand the business: its definitions, its history, and what is happening right now. Data 360, Informatica, and MuleSoft work together to deliver that understanding.

Here’s how the platform comes together:

The Data Layer: Everything starts with the data foundation we just described. This layer combines Data 360, Informatica, and MuleSoft to deliver trusted context. This layer gives Agentforce memory and understanding. It doesn’t just see a “refund request”; it sees the lineage, the real-time status, and also the context that includes the customer profile, allowing it to reason with the clarity of a human expert.

The App Layer: Agents shouldn’t have to relearn your business from scratch. This layer gives them access to 20 years of business logic and workflows already built into Salesforce. Agents can instantly tap into established logic across Sales, Service, Marketing, and Commerce, helping to ensure that every automated action follows your business’s existing rules.

The Agent Layer: This is the command center where you build, govern, and orchestrate your AI agents. With Agentforce, you can deploy specialist agents for specific tasks such as a campaign agent or supply chain agent. Furthermore, with MuleSoft Agent Fabric, you aren’t limited to the Salesforce ecosystem — you can orchestrate and govern third-party agents (like those built on AWS or OpenAI) alongside your native Agentforce agents from a single control plane.

The Experience Layer: Finally, your Agentforce needs to be where your customers and employees are. The experience layer allows you to deploy Agentforce to every employee or customer channel. Whether it’s a voice experience on an 800 number, a chat on your website, or embedded directly inside Slack or ChatGPT, the agent carries its trusted context and security with it.

All four layers reinforce one another, with the data layer providing the trusted context that makes the rest effective. 

An Open and Extensible Foundation: Most critically, these four layers of Agentforce 360 come together as an open and extensible platform, allowing you to integrate with your existing applications and systems. You can ingest data from any source in the data layer, write custom code or automations in the app layer, orchestrate Agentforce and third-part agents together in the agent layer, and embed agents into channels in the experience layer. Agentforce 360 meets you where you are.