Salesforce Einstein aims to democratize AI

CEO Marc Benioff calls it ‘AI for everyone’—but only time will tell how useful the artificial intelligence feature really is

Salesforce Einstein aims to democratize artificial intelligence

Salesforce CEO Marc Benioff unveils Salesforce Einstein at the Dreamforce conference.

Credit: Fredric Paul

Along with virtual reality, artificial intelligence (AI) is perhaps the hottest buzzword in technology these days. AI is turning up everywhere, from chatbots and smart speakers to driverless cars and autonomous everything. And now it’s coming to CRM, courtesy of Salesforce Einstein, which the company says brings machine learning, predictive analytics and natural language processing to the entire Salesforce platform.

At the giant Dreamforce conference in San Francisco this week, Salesforce execs from CEO Marc Benioff on down were predictably enthusiastic about the new features. In a 150-minute keynote address accompanied by and an impressive variety of non-profit and other Salesforce customers, Benioff warned attendees that “you know the world has been changing” and noted that the cloud gives you access to the new world of AI technology.

Artificial intelligence at the push of button

Benioff claimed Salesforce hired more than 100 of the world’s top AI visionaries and data scientists to power this “march to AI.” Salesforce Einstein, he said, was “everybody’s data scientist.”

Salesforce co-founder and CTO Parker Harris added that the goal was to simplify AI the way that Einstein’s famous E=MC2 equation simplified complex interactions. Salesforce wants to bring AI to everyone, he said, so that you wouldn’t have to hire those hard-to-find data scientists. Instead of having to “wrangle” the data you want to use, you just click to add the data you want to work with and Einstein automatically identifies opportunities and makes recommendations for sales, service and marketing—without ever having to tell it what to look for

Einstein gets practical

In sales, for example, Einstein calculates a lead score that sorts leads by who is most likely to convert, making it easier for salespeople to focus their efforts. At Salesforce client Fitbit, Einstein is helping guide customer interactions to create a more cohesive community and provide hints on how users can make the most of their fitness trackers. It even helps create personalized communications and emails designed to help users reach their fitness goals

The idea, Harris said, is for Einstein to help Salesforce customers use AI to change the way they work, making them smarter and enabling them to do their best work.

Do you believe in AI magic?

“I know that sounds like magic,” Benioff acknowledged, “but so did it when we said were going to give the cloud to everyone.”

Yes, Salesforce was clearly right about this whole cloud thing. And the company is obviously not alone in jumping on the AI bandwagon. But I do see some possible pitfalls for Salesforce Einstein, not even counting the slightly unseemly appropriation of the famous scientist’s name and a demeaning caricature of his appearance.

First, it’s too soon to know how well the new feature’s algorithms really work. Do they really identify the best leads to pursue? Do they always find the right touchpoint for a given customer? Even once Einstein is in wide use, it won’t be easy to test how effective it is compared to human intervention. And errors in the algorithms could theoretically send companies down the wrong path without them realizing anything was wrong until after the fact.

Perhaps even more worrysome, it’s also possible that Einstein will be just good enough to enter the “uncanny valley” of CRM, generating interactions that are so close to human ones, they’re almost totally convincing yet remain just different enough to creep people out.

In the end, though, these quibbles may not matter much. No matter what flaws Einstein possesses now, Salesforce will no doubt improve it as time goes on. Competitors, meanwhile, will have their own AI approaches, some of which will bring their own innovations.

Then there’s the question of scale. Even if Einstein is flawed, the AI approach lets it deal with far more situations than could possibly be addressed manually. So even if an AI-powered interaction is not as good as a human touch, it’s still better than nothing.

Must read: Hidden Cause of Slow Internet and how to fix it
View Comments
Join the discussion
Be the first to comment on this article. Our Commenting Policies