Salesforce plans to incorporate generative AI across the platform

Salesforce has been in the news for a lot of non-product reasons of late, so it would be understandable that the company wants to bring the focus back to the business of selling software and away from the machinations of activist investors. Today at the TrailblazerDX developer conference, the company announced a generative AI push with a pilot of technology they are calling "Einstein GPT" that adds ChatGPT-like features across the platform.

“We're announcing Einstein GPT, the world's first generative AI for CRM. I think the future is really bright here. It's creating a tremendous amount of opportunities for innovation within our ecosystem of products as well as our broader ecosystem,” Patrick Stokes, EVP and GM for platform at Salesforce told TechCrunch.

The company has had an underlying intelligence layer, Einstein, which has been running under the hood since 2016, helping bring intelligence across the platform. For example, it could help sales teams find the most likely customer to buy or the ones most likely to churn. The generative AI is more content-oriented and helps business users auto-generate text, pictures and code.

For Salesforce though, the data won’t in most cases be coming from the open internet, as with chat-based search. Instead, it will tie directly into the Salesforce data cloud. That means the new offering not only makes it easier for customers to generate content from their own data, it also provides a way to tie the new technology to a mainstay of the Salesforce platform as a data source.


It also can connect to Flow, the company’s workflow engine, so by adding generative AI to the equation, Salesforce could be helping sell other services, which could increase revenue at a time that it desperately needs it with activist investors pressuring the company to increase profits and reduce costs.

With that kind of cross-platform pollination in mind, Clara Shih, GM at Salesforce, says it will be incorporated across the different pieces of the platform, depending on the job at hand. For Service Cloud, the company will generate a reasonable answer to a customer's inquiry based on information in the knowledge base automatically. Shih says the rep will always have the final say before sharing an AI-generated response with a customer.

“Think about all of the emails and chats that come into service agents today. They get inundated. With Einstein GPT for Service we can auto-generate draft replies so that the agents can respond to customers much faster, and they get final say. They can make any edits before they hit send,” she said.

Sales is also an obvious target for this technology. Imagine entering a question about a company, and the bot finds publicly available information to add or update a company record in the CRM. The AI can also generate an email to this company based on this information, and the salesperson could adjust the letter by asking for more or less formal tone, as one example of interacting with the generative technology.

Gif of Einstein GPT for Sales. A salesperson asks for recent information on a company and it answers the question automatically
Gif of Einstein GPT for Sales. A salesperson asks for recent information on a company and it answers the question automatically

Image Credits: Salesforce

Then there is marketing where employees can quickly generate web copy and pictures for a microsite using generative AI. In Slack, you could summarize long threads to understand them better, or perhaps gain insight on your customers, and ask questions to get back more details from long threads. Finally, coders can work faster using generative AI to create code snippets.

“Einstein GPT for Developers can improve developer productivity with AI-generated code snippets and test cases and comments for languages such as Apex,” Shih said.

She notes that Einstein GPT is built on top of OpenAI's technology, but customers can bring their own AI models or use one that comes out of the box from Salesforce, but given the ties to Salesforce data, and the potential headaches involved in dealing with external models, it seems customers would be more inclined to use Salesforce’s.

It’s not like this announcement came as a surprise to anyone who has been paying attention, given that CEO and Chair Marc Benioff has been telegraphing it for weeks on Twitter that he was enamored with the technology, and Salesforce has historically been known to try and incorporate whatever tech was the popular darling of the moment.

At the same time, the company also announced a $250 million fund under the auspices of its venture investing arm Salesforce Ventures to encourage the growth of the startup ecosystem around generative AI.

The Einstein GPT line of products will be available in pilot starting today with the expectation it will be made generally available some time later this year.