Getty Images/iStockphoto
Salesforce users empowered to build no-code Einstein Copilots
Salesforce admins and developers can build custom generative AI assistants for workflows across the platform for sales, service, marketing and e-commerce.
Let them make bots. That's the message Salesforce sent users today with the release of Einstein 1 Studio, a trio of no-code tools that enable admins and developers to build their own generative AI tools for workflows across the platform.
Included in the initial release are Copilot Builder, which creates GenAI actions that perform tasks; Prompt Builder to insert custom prompts into workflows; and Model Builder, which is the on-ramp for custom data models. The Einstein 1 platform -- not to be confused with last year's AI Cloud, which combines several first-generation Einstein GPT features -- is tightly integrated with Data Cloud and features tools to integrate users' data with generative AI.
Interestingly, while Einstein 1 Studio is generally available, Einstein Copilot is still in beta, the company confirmed. Customers can access Einstein 1 Studio by purchasing Einstein 1 editions or by adding it on to Enterprise or Unlimited editions.
Salesforce's approach to Einstein 1 Studio will be familiar to users, said Rebecca Wettemann, founder of Valoir, an IT research firm. It provides a mix of low-code and no-code tools, much like the platform's predictive analytics AI tools, as well as a more complex set of tools for developers. Some Salesforce users, such as customer service teams, might be ahead of others when it comes to early adoption use cases.
"I think customer service is the obvious one because they already have the data," Wettemann said. "They already have the data in a standard format, whereas a lot of sales folks are looking at it and saying, 'Whoa, I've got to do a lot about my data hygiene in a Salesforce automation solution before I unleash [Einstein Copilots].'"
While Salesforce claims it builds its AI on trust, workers don't trust AI in general yet, according to Valoir's "Language Matters: AI User Perceptions" report released last month. While 84% of workers have experimented with generative AI and 41% indicated they think it could help with search, many respondents question the tools' value versus risk.
Seventeen percent of respondents said they don't believe AI can help them at work at all. Moreover, 50% of respondents said they don't like the word copilot, preferring virtual assistant instead. Only 15% said they think AI will help them jump-start a writing task.
Salesforce bets on owned data
So, what can users trust? Their own data, secured within the platform, Salesforce is betting.
Einstein 1 Studio users can create actions that target information inside a company's Salesforce instance and summarize it. Beta tester Andrew Russo, Salesforce architect for BACA Systems, a robotics manufacturer in Michigan, said security was the key to getting the organization to use Einstein Prompt Builder.
BACA's Einstein Copilots are able to plumb rich, unstructured data in sales and service records to, for example, write personalized emails for salespeople. The company also built a tool to summarize customers' historical service records for technicians as they head out to repair a machine.
"It's not something super magical [that] Salesforce has done, allowing us to securely have our data," Russo said. "We have a lot of data in our org, and we have a lot of emails that are captured. [To] be able to securely understand that is really important to us, because we don't want our data going out to the public internet where other people could access it or train models off of it."
While other major tech vendors offer generative AI copilot tools, such as Microsoft's Copilot for Sales and Copilot for Office 365, Salesforce believes its advantage lies in the data held within its secured platform, said Clara Shih, Salesforce CEO of AI.
"Microsoft has multiple different ones," Shih said. "One Einstein Copilot works across all of Salesforce -- that includes Tableau, MuleSoft [and] Slack too."
Einstein uses natural language conversation, so users are able to have multi-turn conversations with the tool, she said. "And right out of the gate, when you start using Einstein Copilot, it's already knowledgeable and aware of all the data [and] all the business processes in that customer's Salesforce organization," Shih said.
Einstein 1 Studio was released in conjunction with the TrailblazerDX conference, often referred to as TDX, for admins and developers, held March 6-7 in San Francisco. Shih said Salesforce admins, developers and consultants have either earned more than a million AI-related badges on its free technical education site, Trailhead, or the community is very close to passing that milestone.
Don Fluckinger is a senior news writer for TechTarget Editorial. He covers customer experience, digital experience management and end-user computing. Got a tip? Email him.
