Agentforce
This release includes new features such as AI-Assisted Agent Generation, which uses a large language model (LLM) to provide suggested topics, instructions, and actions when creating agents from scratch, the Agentforce Testing Center, which assists companies in testing agents at scale using AI-generated scenarios and expanded language support for Prompt Builder to support a broader range of languages.
Field-based masking in the Einstein Trust Layer is also included in this release, which uses Salesforce field metadata to identify and mask sensitive data using placeholders for better data protection.
Service Cloud
The Spring ’25 release includes new Agentforce Service talents. This edition features Service Cloud Assistant, which generates individualized, multi-step action plans for service representatives, as well as Agentforce for Field Service, an artificial intelligence agent that assists customers in scheduling appointments without requiring them to be transferred to a human.
Sales Cloud
Agentforce for Sales now supports retrieval augmented generation (RAG). With RAG, Agentforce can use a variety of structured and unstructured company and customer data, such as uploaded documents and knowledge articles, to create hyper-personalized responses for lead nurturing. This results in more engaged discussions and speedier contract closings.
Data Cloud
New Salesforce Data Cloud capabilities include Private Connect for Data Cloud, which provides safe, bidirectional data access, Zero Copy File Federation, which allows for big dataset access, and Hybrid Search, which returns more relevant search results from unstructured data. These features work together to improve data access, security, and search capabilities, making Agentforce experiences more reliable and accurate.
Platform
Event Log Objects is a new feature in Platform Event Monitoring that stores log data as regular Salesforce objects, allowing users to readily identify, view, and analyze user behavior.
Agentforce customers can even develop custom agent actions using Event Log Objects and SOQL queries to provide complete summaries of risky user activity. Agentforce can then take rapid action, such as establishing security cases or withdrawing user permissions.