TrailblazerDX - Salesforce taps into ChatGPT for generative AI push across business clouds

(Marc Benioff via Twitter)

We know what the big story is at TrailblazerDX this week. CEO Marc Benioff has scarcely been able to contain his excitement about it on Twitter. But Salesforce today officially launches Einstein GPT, pitched with no knowing understatement as “the world’s first generative AI CRM technology”.

According to the launch collateral, Einstein GPT will enable Salesforce to transform every customer experience with generative AI. It combines Salesforce’s proprietary AI models with generative AI technology from an ecosystem of partners, and real-time data from the Salesforce Data Cloud.

Customers are then able to connect that data to OpenAI’s advanced AI models out of the box or choose their own external model, and use natural-language prompts directly within their Salesforce CRM to generate content that can adapt to changing customer information and needs in real-time.

One example given is that Einstein GPT can generate personalized emails for salespeople to send to customers, generate specific responses for customer service professionals to more quickly answer customer questions, generate targeted content for marketers to increase campaign response rates, and auto-generate code for developers.

Launching today are a number of GPT enabled use cases across the Salesforce cloud line-up:

• Einstein GPT for Sales, which allows users to auto-generate sales tasks, such as composing emails, scheduling meetings, and preparing for the next interaction.
• Einstein GPT for Service, which allows users to generate knowledge articles from past case notes and auto-generate personalized agent chat replies.
• Einstein GPT for Marketing, which allows users to dynamically generate personalized content to engage customers and prospects across email, mobile, web, and advertising
• Einstein GPT for Slack.
• Einstein GPT for Developers, which allows developers to improve productivity with Salesforce Research's proprietary Large Language Model by using an AI chat assistant to generate code and ask questions for languages, such as Apex.

Service implications

Prior to today’s announcement at TrailblazerDX in San Francisco, Clara Shih, General Manager of Salesforce Service Cloud, explained the reason behind the GPT push:

Already we're seeing Large Language models transform a number of disciplines, from drug discovery to language translation. Customer engagement is the next frontier. We believe that the value that generative AI can deliver to enterprises is enormous. But like any new technology, generative AI is not without risks. A trusted approach rooted in business outcomes and ethical design is required.

Seven years ago, we introduced the world to AI purpose-built for CRM. We're so excited now to be leading the industry again, in creating generative AI purpose-built for CRM. Einstein GPT combines Salesforce's proprietary AI models with vetted external generative AI. It's being integrated into every Salesforce cloud, as well as MuleSoft Tableau and Slack, and will transform every sales, service, marketing, commerce and IT experience.

The new offerings are accompanied by an open and extensible ecosystem, Shih said, and will be focused on business outcomes:

Salesforce Einstein GPT will combine our proprietary AI models with vetted partners,  an open ecosystem of vetted partners, including our initial launch partner, OpenAI. This means that customers have choice. Customers can bring their own generative AI models, or choose one of our 'out-of-the-box' generative AI options from vetted partners.

AI is only as valuable as the data and workflows. Today, hundreds of thousands of companies, and millions of employees at these companies, use Salesforce for their their sales workflows or service workflows, marketing, commerce and IT workflows. We've got the relevant context to really make the most out of generative AI. In addition, our customers customers are continuously changing. Our Salesforce Data Cloud means that we can allow the Generative AI to continuously adapt as customer information and needs change.

Last but not least, trust. With great power comes great responsibility. And trust begins with data. Starting with our our secure public cloud architecture, HyperForce, we can ensure secure data access, that personally identifiable information is protected and that we're complying with local and regional data residency requirements from a regulatory and compliance perspective.

In addition, we've partnered closely across our product and engineering teams in our Office for Ethical and Humane Use of Technology to ensure that we're building in ethical guardrails throughout Einstein GPT, like we've done for many years across the rest of our technologies. A great example of this is that we are we have designed into Einstein GPT, a human in the loop, which means that an employee - whether that's a seller, a marketing manager, or a customer service agent - has final say and can make any edits on what Generative AI creates as a draft, before hitting send or submit.

Narrowing down on the customer service implications, Shih added:

[Service agents] are inundated today with chats and emails that they have to respond to. Einstein GPT for Service will help them auto-draft relevant tailored responses to reduce the average handle time and get the answer right the first time. In addition, we will help take agent case notes that are created today and turn those into drafts of new knowledge articles. Today, on average, it takes 10 days for a new knowledge article to be created across industries. We believe that generative AI and Einstein GPT will cut this down to a fraction.

But given that most studies suggest that customers want to talk to a human being when engaging with customer service functions, not a chatbot, how does Einstein GPT make a difference? According to Shih:

The lines are really blurring between chatbot interactions and human interactions. I think one of the most exciting areas is when you're using chatbots for very simple FAQs and also to collect customer context that can be passed off to a service agent for the higher order problem solving. So, I don't see them as mutually exclusive. They really help each other it's human-plus-AI together.

My take

It’s a big deal for Salesforce. Phil Wainewright is on the ground in San Francisco at TrailblazerDX. I’ll leave it to him to give the full skinny as the week plays out.