Salesforce Inc. today is jumping aboard the generative artificial intelligence train with the addition of large language model support across its portfolio of applications.

Generative AI can now be incorporated into the Einstein artificial intelligence platform, Salesforce Data Cloud, Tableau analytics, Flow process manager and MuleSoft integration toolset for queries, code generation and business process automation assistance.

Generative AI has been popularized by last fall’s release of ChatGPT by OpenAI LP. In addition to responding to queries in natural language, ChatGPT can to generate and analyze code as well as extract insights from raw data.

Trusted and extendable

An important feature of the new version of the Salesforce AI Cloud, which is also known as Einstein, is the Einstein Trust Layer, which the company said makes its cloud platform open and extensible to allow customers to choose their preferred LLMs from Salesforce and other companies. The Trust Layer is said to prevent proprietary data from becoming part of public models, improve the quality of AI-generated content and integrate generative AI responses into business processes while ensuring that data privacy, security, residency and compliance regulations are met.

Salesforce highlighted the trustworthiness of AI models as a major factor in the development of Trust Layer, citing its own research that found that 73% of employees believe generative AI introduces new security risks. Nearly 60% of those who plan to use the technology said they don’t know how to keep data secure.

In addition to Salesforce’s own LLMs, the AI Cloud also supports models from OpenAI, Anthropic PBC and Cohere Inc. with additional support for other third-party models planned. Einstein GPT also includes a gateway that can be used to incorporate models from Amazon Web Services Inc. and Google LLCs cloud as well as domain-specific and on-premises hosted models.

Customer prompts and responses never leave the Salesforce infrastructure and customers that have trained their own models can connect directly to the AI Cloud through the Trust Layer. Salesforce said it has also developed optimized AI prompts that can be customized to the customer company.

Many uses

The company offered examples of how generative pretrained transformer technology can be incorporated into specific use cases. Salespeople can automatically generate personalized emails informed by customer relationship management data. Service teams can create service briefings, case summaries and work orders based on case data and customer histories.

Marketers can segment audiences using natural language queries and e-commerce professionals can tailor product descriptions to buyers based on stored customer data.

Integration with Salesforce’s Slack collaboration platform allows users to build new prompted no-code workflows. Sales managers can generate visualizations in Tableau with natural-language prompts and developers can get intelligent code suggestions and completions directly within Visual Studio.

In related announcements, the company said its Salesforce Ventures Generative AI Fund will be doubled in size to $500 million and that new startups will be added. The company also introduced the Salesforce AI for Impact Accelerator, a philanthropic initiative that provides funding, pro-bono expertise and technology to nonprofit and “purpose-driven” organizations.

Photo: Salesforce