Salesforce introduces Enterprise AI Harness foundation

Salesforce Inc. recently unveiled two new tools aimed at helping businesses build and manage artificial intelligence (AI) agents. These tools are part of the company's Enterprise AI Harness, which focuses on enhancing the reliability and security of AI agents. The first offering introduced by Salesforce is the Enterprise AI Harness, a suite of technologies designed to improve the performance and security of AI agents.

The Enterprise AI Harness is composed of six main feature collections. Half of these focus on the business data used by AI agents to complete tasks. One feature, Trusted Context, integrates various data assets such as customer records and product availability information. Salesforce also plans to incorporate support for "real-time signals," which refer to buyer actions that necessitate a quick response from the company.

Another set of features within the Enterprise AI Harness is dedicated to ensuring data accuracy. The company intends to provide multiple information quality controls, including a data lineage tool. Additionally, cybersecurity teams will have the capability to regulate which agent can access which dataset and under what conditions.

The Enterprise AI Harness also includes an "intelligent model routing" mechanism. This feature automatically directs agent requests to the most suitable large language model (LLM) based on factors like accuracy and cost. This mechanism is particularly useful in AI agent projects where multiple LLMs are used to optimize costs. Some agents handle complex tasks using a frontier model, while simpler agents make requests to a more affordable LLM.

Salesforce emphasizes that Enterprise AI Harness will enable customers to pair their AI agents with deterministic reliability controls, ensuring consistent responses to prompts. Furthermore, the offering aims to simplify the process of connecting agents to external applications.

The second offering previewed by Salesforce is the AI Control Plane. This tool allows administrators to centrally monitor the performance and inference costs of AI agents. Additionally, it helps ensure that agents comply with cybersecurity policies. As Salesforce President and Chief Platform and Engineering Officer Rohan Kumar stated, "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." Salesforce's Enterprise AI Harness is designed to address these challenges.