Clara Shih, CEO of Salesforce AI, describes Einstein Copilot as a “new conversational AI assistant for every Salesforce customer.” She also notes that the development team has been working tirelessly to bring this new category of AI technology to market quickly.
To further validate the need for Einstein Copilot, Salesforce has released new research from its Slack business unit. The research shows that 86% of IT executives expect generative AI to have a significant impact on their enterprises. Additionally, 80% of employees who are already using generative AI tools report a boost in productivity.
While Einstein Copilot is an exciting addition to Salesforce’s AI capabilities, it’s important to note that the company’s original Einstein technology predated the generative AI era. The platform, which was built on predictive machine learning and AI algorithms, has been serving over 80 billion predictions per day since 2020. In March 2023, Salesforce expanded the platform with the debut of Einstein GPT, which introduced predefined AI use cases embedded in specific Salesforce workflows.
Despite the gen AI craze, Shih emphasizes that the predictive AI capabilities of Einstein remain a cornerstone of Salesforce’s AI offerings. These capabilities enable features such as providing salespeople with recommendations for the next best action or helping sales managers forecast sales quotas. Einstein Copilot takes this a step further by offering an open-ended assistant that can answer any question across Salesforce data and workflows.
The foundation for Einstein Copilot lies in Salesforce’s Hyperforce data architecture. This architecture provides data residency and compliance for enterprise use cases. On top of Hyperforce is the Salesforce Data Cloud, which unifies and cleanses data from multiple sources. The Einstein LLM (large language model) gateway sits on top of the data stack, pulling in various gen AI models depending on the use case. Salesforce has developed its own LLMs and has partnerships with leading vendors like OpenAI, Google, Anthropic, and Cohere.
To ensure data privacy, security, and trustworthiness, Salesforce has implemented the Einstein Trust Layer. This layer provides toxicity filtering, citations, and other tools necessary for enterprise consumption of AI. Collectively, these underlying technologies are branded as Einstein 1, forming the overall AI platform and strategy at Salesforce.
While other organizations are also developing AI copilots and conversational UI interfaces, Shih highlights the importance of context in Salesforce’s approach. The metadata model used by Salesforce provides structured data that gives context to the information being analyzed. This metadata allows AI to function properly and enables Salesforce’s conversational AI assistant to understand the nuances of different data sets and workflows.
In conclusion, Salesforce’s release of Einstein Copilot marks a significant advancement in conversational AI technology. By providing a more open and intuitive interface for interacting with data and workflows, this new assistant empowers Salesforce users to make better-informed decisions and improve productivity. With its strong foundation in predictive AI and its focus on context through metadata, Einstein Copilot sets itself apart from other AI copilots in the market. As Salesforce continues to innovate in the AI space, users can expect even greater advancements in the future.
