Salesforce’s Einstein Finds Its Voice


Salesforce showcased on Wednesday new natural language features for enabling its Einstein AI platform to further simplify the work of sales agents, marketers and business leaders.

On 'AI Day' at the Dreamforce conference, members of Salesforce's data science team demonstrated how Einstein Voice—an emerging speech interface—can be used to better access sales figures and projections, automate repetitive tasks, and receive business insights, including through Amazon's and Google's smart speakers.

Jim Sinai, Salesforce's vice president of marketing, said in a keynote that Salesforce had been working on Einstein Voice for the last year to better deliver Einstein's data discovery, deep learning and machine learning capabilities.

"AI is about helping all of us live our maximum potential and work to our maximum potential," Sinai said.

Sponsored post

[Related: Salesforce's Benioff Paints Exuberant Vision of ‘Inclusive Capitalism’]

Richard Socher, Salesforce's chief scientist, told Dreamforce attendees Einstein Voice launches with an initial focus on three use cases.

First is the ability to conversationally log notes, a free feature. Manually entering notes "is probably one of the least-liked features of the CRM," Socher told attendees.

Salesforce was also looking to take advantage of the 50 million smart speakers that are deployed just in the U.S.

"We want to be able to connect Salesforce through that kind of experience," Socher said.

Salesforce built integrations for Einstein Voice with Amazon Alexa and Google Home, enabling sales professionals to get personalized daily briefings by asking those devices.

Finally, Einstein Voice will power voice-driven dashboards that present continually updated analytics.

Those developments are "just a step one," Socher said.

Natural language recognition has become largely commoditized, he told attendees, with products like Apple's Siri now commonly used to accept voice commands or transcribe text.

"Where it gets hard and interesting is the natural language processing behind that," Socher said.

Einstein's processing capabilities allow Voice to do things like learn the jargon and unique vocabularies of customers and extract from spoken queries the names of relevant people and companies.

"Then we integrate all of that into the CRM," Socher said.

Salesforce customers can pass on Voice's capabilities to their own customers, he said, allowing them to deliver custom apps that work with the smart speakers.

Einstein Voice complements an emerging set of capabilities: Einstein Analytics for forecasting, Einstein Opportunity for scoring leads, and Einstein Prediction Builder, in beta, for crafting an intelligent rules engine that suggests strategy through next-best action.

There's also Einstein Vision, an image recognition tool that allows users to take a picture that can be compared to a catalog to identify a product.

Salesforce has also used Voice to update Einstein Bots, voice enabling the chatbot builder released earlier in the year.

"Einstein Voice represents a long-awaited upshift in natural language processing technology that will be an immediate competitive differentiator for our clients that support large-scale field sales and service operations," said Jill Harrison, vice president of Silverline Ventures, a Salesforce partner based in New York.

Partners like Silverline will build the next generation of mobile apps powered by the new voice-enabled service.

"Productivity is king for these highly mobile and territory-driven professionals," she told CRN.