Salesforce Unveils Composable Architecture for Trustworthy Enterprise AI
Salesforce has introduced the Trusted Enterprise AI Harness, a composable architecture for enterprises that combines six trusted capabilities and an AI Control Plane. The announcement frames this as a response to the challenges of the Agentic Enterprise.
The Agentic Enterprise is characterized by agents taking on complex work, understanding what's happening, deciding what to do next, and working alongside people and other agents. However, equipping them to do this work reliably, securely, and at scale without building and managing capabilities separately for every agent or AI experience has become a challenge.
Salesforce illustrates the gap with an example: 'Can we fulfill this order today?' The company notes that no single system holds the complete answer. CRM knows the customer and relationship, ERP knows inventory and fulfillment, contracts contain commitments and entitlements, analytics provides business definitions, policies determine what can be promised, previous interactions supply memory, and business processes determine how work should get done.
The six trusted capabilities of the Trusted Enterprise AI Harness are: Trusted Context, which brings together customer context with data, metadata, semantics, knowledge, real-time signals, memory, and an understanding of how work gets done; Trusted Agency, which supplies reasoning, planning, state, memory, collaboration, and orchestration agents need to pursue complex business outcomes; Trusted Action, which securely connects AI to applications, APIs, workflows, tools, and business processes; Trusted Governance, which covers the data, metadata, policies, and processes AI relies on; Trusted Security, which applies identity, permissions, privacy, data protection, and runtime security to what AI can access and what agents are allowed to do; and Trusted Models, which lets enterprises securely connect the right model for each job.
Rohan Kumar, Salesforce's President, Chief Platform and Engineering Officer, said that 'The Agentic Enterprise won't be defined by which model a company chooses.' Instead, it will be differentiated by the trusted, proprietary context it brings to intelligence, starting with the customer, and its ability to securely turn that context into action.
The new AI Control Plane gives businesses one place to see, manage, and control agents and AI across the enterprise. It enables companies to discover and register agents and AI capabilities, establish identity and policy, manage lifecycle, evaluate performance, observe behavior and outcomes, and control cost, across both Salesforce and third-party AI.