Salesforce Introduces the Trusted Enterprise AI Harness

Salesforce introduces the Enterprise AI Harness, a new architecture designed to provide AI agents with a shared understanding of customers and business operations, enabling them to act reliably and securely at scale. The system integrates six capabilities—context, agency, action, governance, security, and models—into a composable framework built on existing customer data, processes, and controls. Businesses can adopt the full suite or select components, combining Salesforce technology with third-party tools or their own systems.

The framework includes a new AI Control Plane that offers centralized visibility and management of AI agents across the enterprise, allowing companies to track performance, enforce policies, and control costs. This addresses the challenge of maintaining oversight as AI agents proliferate across different business functions and systems. The Control Plane supports discovery, registration, identity management, lifecycle oversight, and performance evaluation.

The Enterprise AI Harness breaks down AI decision-making into six trusted components: Trusted Context, Trusted Agency, Trusted Action, Trusted Governance, Trusted Security, and Trusted Models. Each component ensures AI operates with accurate data, follows business rules, executes securely, and adapts to evolving model requirements. For example, Trusted Context combines customer data with real-time signals to determine order fulfillment feasibility.

The architecture leverages Salesforce’s existing technologies, including Data 360, Informatica, MuleSoft, Agent Fabric, Tableau, Agentforce, Salesforce Guardian, and the Salesforce Platform. It is designed to enhance proprietary enterprise context—such as customer relationships and business processes—over time, creating a durable source of differentiation as AI capabilities evolve.