Salesforce has unveiled Einstein, a deep learning platform it hopes will form the foundation of all of its future SaaS offerings.
Touting years of development and hundreds of millions of dollars in both in-house development and acquisitions, Einstein will be baked into pretty much every major product Salesforce offers. Salesforce is rolling out Sales, Service, Marketing, Commerce, Community, Analytics and IoT packages that will include the Einstein AI tools, in some cases for free and in others as charged add-ons.
More ReadingIf someone can translate Salesforce's $700m cool Krux plan into English, we'd be really gratefulIn the name of love, don't scalp your U2 aka Dreamfest tickets, warns SalesforceSalesforce asks Europe to stop Microsoft buying LinkedIn after itself trying to buy LinkedInLack of Hurd mentality at Oracle OpenWorld: Co-CEO's cloud claims fall flat live on stageSpoiler alert: What Oracle is going to announce today
The roles played by Einstein will vary by service, but in most cases the aim is to add predictive and analytics capabilities to existing Salesforce services.
One example is Sales, where Einstein will be used to help predict when a lead has a better chance of turning into a sale, or alerting salespeople when a contact might be considering a competitor.
In Marketing applications, Einstein tools could be used to analyze images posted through social media or gauge the sentiment of customers who use products and suggest where to target marketing.
For Salesforce, the aim of Einstein is to get AI capabilities to as many of its customers in as many different forms as possible.
"For the vast majority of companies, [AI] is too hard, they can't apply it to their business processes," said Salesforce senior VP of product strategy John Ball.
"This is democratizing AI so that every company can benefit."
That diversity, however, could also be a shortcoming for Einstein. AI and deep learning are technologies still facing a long road in their development, and being able to implement them reliably in one task is difficult enough, let alone throughout more than a half dozen different services.
Salesforce execs noted that the effectiveness of the deep learning tools will vary based on the amount of data offered to them, and customers will be able to choose whether they allow Einstein to use their (anonymized) data to develop those analysis tools. Those who opt to give more data to Einstein and use the tools for a longer period would then get better results.
How useful those results will be, and in how many cases, remains to be seen. Some companies may find the analytics tools offer little in the way of a competitive edge or useful insight, but rather point out the obvious and unnecessary (much like a certain unloved Redmond personal assistant). Other customers simply may not be large enough or have a huge data set that would benefit from the AI features.
The timing of the announcement is no coincidence, either. Pushed out just hours before Larry Ellison is scheduled to take the stage at Oracle OpenWorld, Einstein is Salesforce's bid to distance itself from the encroaching hordes of enterprise giants looking to usurp its cloud CRM throne.
Having an AI layer that contributes throughout its product lines would once again allow Salesforce to get out ahead of the enterprise companies that it left in the dust with cloud computing a decade ago. That may, however, be asking too much of a highly complex platform still very much a work in progress. ®