At its European analyst day in London, Salesforce focused on its recent product announcements surrounding Einstein, its foray into artificial intelligence (AI). Two things make Einstein particularly interesting: the platform approach, which creates the possibility of infusing AI into each of the vendor's cloud solutions; and what Salesforce believes are its differentiating features – its customers' large data volumes, and application integration. While still relatively early days, Einstein and the capabilities it represents – intelligence automation and embedded analytics – have real potential upside for enterprises.
Einstein is woven into Salesforce clouds, but its success will depend on the data
Salesforce announced Einstein at its 2016 Dreamforce event, and has spent 2017 letting customers know what Einstein can help them achieve. At its heart, Einstein is a platform that combines multiple AI technologies acquired and developed by Salesforce that surface in its cloud applications through insights delivered against specific processes, for example, prioritizing leads. It would be fair to say that much of what constitutes Einstein is a "black box." Should that matter to users of the technology? I strongly argue no. In fact, I argue this is an extension of a principle I urge adherence to: the best technology is the one you don't know you're using.
Since the initial announcement, Einstein has been put to use in easy-to-recognize ways in Salesforce's clouds. Mentioned already was predictive lead scoring; other examples include recommended case classification in Service Cloud, product recommendations in the Commerce Cloud, and predictive capabilities added to Wave apps. In each of these cases, although the technology is founded in a platform, its consumption is made easy by offering insight in a well-established process and integrated directly into the user's everyday application. This is the tip of the iceberg, as Ovum expects Salesforce to make Einstein an integral part of every aspect of its portfolio.
Adoption of AI capabilities presents both technological and business challenges. From a technology perspective, the Einstein platform is largely only as good as the data it ingests and learns from. The obvious side effect is that if that data is either limited in scale or detail, or of poor quality, Einstein's automated insights will be limited. From the business perspective, two things stand out: Einstein is – for some functionality – separately chargeable for Salesforce customers; and it is still unknown how long it will take those same customers to "trust" the insights served by Einstein. Trust is an evolution, and closely linked to the data question; in both cases, it's a journey and one that will not be completed overnight. Thinking about the cost, I do believe Salesforce (and others) can charge additionally for these capabilities; whether they can continue to do so will depend on the success of deployments, and – pivotally – clear documentation of those benefits as proof points.
Tom M. Pringle, Head of Applications Research, Ovum