Salesforce Unveils Tools for Secure AI Agent Management
Salesforce has announced two new offerings to help customers build and manage AI agents. The Enterprise AI Harness is designed to improve the reliability and security of AI agents, while the AI Control Plane will enable administrators to centrally monitor AI agent performance and inference costs.
The Enterprise AI Harness includes six main feature collections, with half focusing on business data used by AI agents. This includes features such as Trusted Context, which brings together customer records and product availability information, as well as support for real-time signals.
Salesforce has also included an 'intelligent model routing' mechanism in the Enterprise AI Harness, designed to ease the use case of using multiple large language models (LLMs) in AI agent projects. This will automatically send agent requests to the most suitable LLM based on parameters such as accuracy and cost.
Additionally, Salesforce has included deterministic reliability controls in the Enterprise AI Harness, which will enable customers to pair their AI agents with these controls to ensure consistent results. The company also plans to provide multiple information quality controls, including a data lineage tool.