Salesforce has unveiled the introduction of AI Cloud, a collection of products designed to enhance CRM productivity in enterprises by leveraging Salesforce’s AI applications. Scheduled for release this year, the AI Cloud platform provides hosting and delivery services for a variety of AI models, with a particular focus on text generation. It collaborates with partners such as Amazon Web Services, Anthropic, Cohere, and OpenAI, utilizing Salesforce’s cloud infrastructure.
During the announcement in New York City, Marc Benioff, Salesforce’s co-founder and CEO, emphasized the significance of generative AI, stating that it may be the most important technology of our time.
In addition to the partner models, customers have access to first-party models developed by Salesforce’s AI research division, enabling features like code generation and business process automation. Customers also have the option to bring their own custom-trained models to the platform, while retaining their data on their own infrastructure.
Salesforce’s flagship products, including Data Cloud, Tableau, Flow, and MuleSoft, benefit from the advanced capabilities provided by the AI models within AI Cloud. These models encompass Sales GPT, Service GPT, Marketing GPT, Commerce GPT, Slack GPT, Tableau GPT, Flow GPT, and Apex GPT, totaling nine models.
However, AI Cloud currently lacks an image-generation model similar to DALL-E 2 and Stable Diffusion. Salesforce is actively working on developing such a model, aiming to address concerns such as copyright and toxicity before making it available to the public.
Sales GPT facilitates the rapid creation of personalized emails, while Service GPT assists in generating service briefings, case summaries, and work orders based on case data and customer history. Marketing GPT and Commerce GPT generate audience segments for targeted marketing and tailor product descriptions to individual buyers using customer data, offering recommendations to increase average order value.
The remaining models—Slack GPT, Tableau GPT, Flow GPT, and Apex GPT—have more specialized applications. Slack GPT and Flow GPT allow users to create no-code workflows incorporating AI actions within Slack or Flow. Tableau GPT generates visualizations based on natural language queries and uncovers data insights. Apex GPT scans for code vulnerabilities and suggests inline code for Salesforce’s proprietary programming language, Apex.
Several models, such as Slack GPT, Commerce GPT, Sales GPT, and Service GPT, are already operational. The remaining models, except for Flow GPT, which will be available in October, are scheduled to be launched as early as this month.
This move by Salesforce is reminiscent of Amazon’s launch of Bedrock, which offers a range of models trained in-house by AWS, along with pre-trained models for startup partners.
Furthermore, Salesforce has partnered with Google Cloud to expand their strategic partnership, aiming to help businesses leverage AI and data to deliver more personalized customer experiences.
To ensure the privacy and security of these models, similar to NeMo Guardrails, Salesforce has introduced the Einstein Trust Layer within AI Cloud. This layer is designed to moderate content while preserving the privacy of user-sensitive data. It provides openness and extensibility, allowing customers to use third-party LLMs and bring their own models (BYOM) to the Salesforce platform. The Einstein GPT Trust Layer will be generally available starting this month.
To differentiate AI Cloud from other managed AI service providers, Salesforce is introducing a range of prompt “templates” and tools for building prompt templates. These templates utilize “optimized” AI prompts that leverage “harmonized” data to ensure that the generated content aligns with a company’s specific requirements. By grounding the model-generated outputs in context, these prompts significantly enhance the quality and relevance of the generated content.