Salesforce Unveils Enterprise AI Harness and Control Plane for Agent Management

Salesforce Inc. has revealed a pair of solutions aimed at enabling customers to develop and oversee AI agents. Creating an agent from a large language model involves augmenting it with various components—such as prompts, database connectors, and other technical elements. Collectively, these customization components form what is referred to as a harness.

The first solution unveiled by Salesforce carries the name Enterprise AI Harness. This platform brings together a suite of capabilities intended to strengthen the trustworthiness and security posture of customer-deployed AI agents. According to Salesforce, several underlying technologies are currently accessible through its existing cloud offerings, while additional capabilities will become available starting in early fiscal year 2028, beginning in February.

Enterprise AI Harness will be organized around six primary feature groupings, with three of them concentrating on the business information that powers AI agent operations.

Data Management and Quality

One feature set, termed Trusted Context, consolidates various data resources including customer information and inventory details. The platform will incorporate handling for real-time signals—indicators of buyer behavior that prompt organizations to modify their engagement strategies quickly.

A separate collection of Enterprise AI Harness features addresses data accuracy for AI agent operations. Salesforce intends to supply various data quality mechanisms, including a tool for tracking data lineage. Security personnel will receive capabilities to establish restrictions on which agents can access particular datasets and in what manner.

Model Routing and Reliability

Organizations frequently deploy multiple LLMs across AI agent implementations to balance performance with expenditure. Sophisticated tasks rely on a frontier model, while routine operations route to smaller, more economical alternatives. Enterprise AI Harness incorporates an intelligent model routing feature that directs agent requests to the most fitting LLM according to factors like precision and expense.

Because LLMs produce variable outputs—identical inputs can yield different results across separate runs—Salesforce designed Enterprise AI Harness to pair agents with deterministic reliability mechanisms. The platform also streamlines the process of linking agents to third-party systems.

AI Control Plane

Alongside Enterprise AI Harness, Salesforce introduced the AI Control Plane, a management solution enabling administrators to observe agent performance and track inference expenditures from a unified location. The offering further assists organizations in maintaining agent compliance with security requirements.

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. That's what we're building with Salesforce's Enterprise AI Harness.

Rohan Kumar, Salesforce president and chief platform and engineering officer