Today, during the Dreamforce customer conference in San Francisco, Salesforce unveiled Einstein Copilot Studio, a tool that empowers customers to tailor the Salesforce base Einstein GPT and Einstein Copilot offerings to their unique needs.
Einstein Copilot Studio is composed of three core components, as described by Clara Shih, CEO of Salesforce AI:
Prompt Builder: This first element allows customers to customize prompt templates within Einstein GPT. This means users can incorporate their own custom prompts tailored to their products, brand, or specific business requirements not covered by the default Einstein GPT or Einstein Copilot offerings.
Skills Builder: The Skills Builder component enables companies to add actions to prompts. It goes beyond data access and question answering, allowing businesses to control and specify which workflows Einstein Copilot can access and execute. Examples include tasks like competitor analysis or objection handling.
Bring Your Own Model: The final piece of Einstein Copilot Studio permits users to integrate their own models or utilize supported third-party offerings such as Anthropic, Cohere, Databricks, Google Cloud’s Vertex AI, and OpenAI. This feature is designed for larger enterprises with robust data science teams who want to bring their proprietary models into the system.
Brent Leary, founder and principal analyst at CRM Essentials, highlights that Studio addresses the needs of companies requiring more advanced tooling. It serves as a centralized platform accessible across Salesforce’s various offerings, streamlining prompt and skills creation for administrators to meet specific organizational requirements.
Salesforce acknowledges the challenges associated with large language models, such as hallucinations, biases, and inappropriate responses. To address these issues, they are developing the “Einstein Trust Layer,” a system that assists companies in managing security, governance, and data privacy. Leveraging data from the Data Cloud (formerly known as Genie) introduced at the previous Dreamforce, Salesforce aims to provide accurate answers based on information in its databases.
Customers using Einstein Copilot Studio can fine-tune prompts and skills based on the nature of the question and the level of risk involved. This includes deciding whether to fully automate a response or involve a human in certain cases, ensuring a responsible and secure approach to AI assistance.
Clara Shih acknowledges that while AI may impact certain jobs, it also creates new opportunities, such as the role of a prompt engineer.
Einstein Copilot Studio is set to enter pilot testing in the upcoming fall, while the Einstein Trust Layer will become generally available across the Einstein platform next month, although specific dates were not provided by Salesforce.
