Salesforce has announced AI Cloud, a suite of enterprise solutions using generative artificial intelligence (AI).
The primary function of AI Cloud is to increase productivity in customer experiences and enterprise workflows, says Salesforce. A key component of the AI Cloud is the Einstein GPT Trust Layer, designed to manage risks associated with generative AI, such as data security and compliance. Einstein, Salesforce’s AI for CRM, processes over a trillion predictions weekly across Salesforce applications
AI Cloud promises to deliver several practical applications, according to Salesforce. It can help sales representatives create personalized emails automatically. For service teams, it offers auto-generated chat responses and case summaries. Marketers can benefit from auto-generated content to reach out to customers and prospects, while commerce teams receive auto-generated insights and recommendations for customized experiences. Developers can also benefit from AI Cloud with features such as auto-generation of code, bug predictions, and suggested fixes.
However, the adoption of generative AI comes with risks, such as hallucinations, privacy concerns, toxicity, bias, and data governance issues. Salesforce’s research showed that 73% of employees believe generative AI presents new security risks, and 60% of prospective users are unsure of how to keep their data secure. The Einstein GPT Trust Layer in AI Cloud aims to bridge this trust gap by preventing large language models from retaining sensitive customer data, says Salesforce.
AI Cloud integrates several Salesforce technologies, including Einstein, Data Cloud, Tableau, Flow, and MuleSoft. The suite offers an open ecosystem, with Salesforce hosting large language models from other companies such as Amazon Web Services (AWS), Anthropic, and Cohere, as well. This hosting occurs within Salesforce’s infrastructure, maintaining data prompts and responses in a secure environment.
Salesforce will offer its own large language models through AI Cloud, including CodeGen, CodeT5+, and CodeTF. These models provide advanced capabilities like code generation and business process automation assistance.
AI Cloud also allows customers to use their own domain-specific models trained outside of Salesforce, while storing data on their infrastructure. This function is available through the Einstein GPT Trust Layer, and customer data can remain within the customers’ trust boundaries.
[Image courtesy: Salesforce]