Salesforce Formalizes the Agent Governance Stack Into a Single Platform

Salesforce announced its Trusted Enterprise AI Harness on September 10, 2026, a six-capability agent governance platform built around a new AI Control Plane that can discover, register, and manage both Salesforce and third-party AI agents. Salesforce President and Chief Platform and Engineering Officer Rohan Kumar said the differentiator for enterprises will be "the trusted, proprietary context it brings to that intelligence," not the model chosen. The Harness is not scheduled for general availability until early fiscal year 2028, which begins in February 2027, and pricing has not been disclosed.

Most enterprise AI initiatives are currently stuck in a cycle of pilot projects that never reach production. Data from IDC and Lenovo suggests that 88% of enterprises with agent initiatives never ship to production https://venturebeat.com/data-infrastructure/survey-reveals-how-enterprises-are-really-deploying-ai-agents/ , while Gartner reports that over 40% of agentic AI projects are cancelled due to escalating costs and inadequate risk controls. For decision-makers, the promise of autonomous agents is colliding with the reality of operational chaos. Salesforce is attempting to solve this by formalizing the agent governance /glossary/agent-governance/ stack. On September 10, 2026, the company announced its Trusted Enterprise AI Harness https://www.salesforce.com/agentforce/ . This is a strategic move to position Salesforce as the central management layer for agentic AI /learn/what-is-agentic-ai/ across the entire enterprise, not just within its own ecosystem. The Harness consists of six core capabilities: Trusted Context, Trusted Agency, Trusted Action, Trusted Governance, Trusted Security, and Trusted Models. Central to this is a new AI Control Plane, which allows organizations to discover and register agents, establish identity and policy, manage lifecycles, evaluate performance, and control costs. Crucially, this control plane is designed to manage both Salesforce and third-party AI agents. Rohan Kumar, Salesforce’s President and Chief Platform and Engineering Officer, argues that the competitive edge for companies will not be the model they choose, but the proprietary context they apply. As Kumar stated, “The Agentic Enterprise won’t be defined by which model a company chooses. Models will continue to change, and intelligence will increasingly be available everywhere. What will differentiate an enterprise is the trusted, proprietary context it brings to that intelligence – starting with the customer – and its ability to securely turn that context into action.” This announcement follows the recent Claudeforce /claudeforce-sql-to-agentforce-the-7b-logic-layer-that-splits-the-ai-model-wars-in-half/ initiative, where Salesforce deepened its partnership with Anthropic to make Claude the default reasoning engine across its platform. By building both the reasoning engine layer and the new governance layer, Salesforce is attempting to capture the full stack of agentic operations. The market need for this is clear. A VentureBeat Intelligence survey from July 2026 https://venturebeat.com/data-infrastructure/survey-reveals-how-enterprises-are-really-deploying-ai-agents/ found that 85% of enterprises run two or more agent orchestration platforms, with an average of 3.1 platforms per company. This fragmentation is exactly what Salesforce is targeting. By offering a centralized control plane, they are betting that enterprises will pay to consolidate this sprawl. However, this creates a tension for buyers. While Salesforce promises an open architecture – supporting MCP, APIs, Skills, and Plug-ins – the goal is clearly to own the management layer for all agents. Shawn Malhotra, CTO of Rocket Mortgage, noted the importance of this flexibility: “AI is moving faster than any technology we’ve seen – which is exactly why composability matters so much to us. We don’t want to bet our future on one closed stack; we want the freedom to adapt as the landscape shifts. That’s what resonated with us about Salesforce’s new Enterprise AI Harness.” The challenge for decision-makers is balancing this promise of composability against the risk of vendor lock-in. If you centralize your agent governance on a single platform, you are effectively betting on that platform’s ability to remain neutral and interoperable as the market evolves. There is also a significant wait ahead. The Harness is not scheduled for general availability until early fiscal year 2028, which begins in February 2027. Pricing details have not been disclosed. This announcement serves as a strategic positioning move ahead of Dreamforce, setting the stage for how Salesforce intends to compete in the agentic enterprise market. For now, enterprise buyers should watch how Salesforce integrates third-party agents into the Control Plane. The success of this initiative depends on whether it can actually reduce the production gap or if it simply adds another layer of complexity to an already fragmented stack.