ENTER SALESFORCE’S EINSTEIN
Salesforce.com (market cap = $116 billion) has just announced the release of Einstein Voice enabled AI, the smartphone ready equivalent to Alexa without the need for specialized hardware. Salesforce has long been in the business of accepting and digesting CRM data – mostly through traditional tools like keyboards – but has increasingly relied upon voice-enabled services to support daily workforce operations. This Einstein release is the expansion of all that groundwork. From the TechCrunch article, “Einstein Voice gives Salesforce users gift of gab”:
“…you can conversationally enter information about a meeting, get daily briefings about key information on your day’s meetings (particularly nice for salespeople who spend their day in the car) and interact with Salesforce data dashboards by asking questions instead of typing queries.”
With this update to its existing services, Salesforce has an opportunity to fine tune its own AI towards capturing the business voice-enablement market coveted by enterprise versions of voice enabled platforms (Alexa, Google Assistant). The company also holds advantageous positions – being part of the installed base – in many industries. IT has already committed the resources for subscriptions and support; this release just adds to the existing line items.
So Einstein is a full-fledged competitor to the popular voice-enabled platforms. The next question is .. how effective is Einstein?
Most voice-enabled platforms struggle to capture – accurately – open ended text. This capability is arguably the holy grail of voice-enabled experiences. Einstein purports to do open-ended dictation, a weakness for Alexa (and to a lesser extent Google Assistant devices. Of course, Einstein has the natural advantage of operating in a captured domain where the conversational context is focused and limited. Einstein can rely on users to use specific vocabulary and grammar in predictable ways, a headstart towards voice recognition accuracy.
Einstein has another advantage. Voice-training is helpful towards improving recognition accuracy. Salesforce has had months of days-long training with users in preparation for this release. It was a nicely contained testing ground which provided a lot of data upon which to draw inferences and make adjustments. They knew beforehand what type of information to expect and what the AI perceived, which would be of ready use in the field for on-the-fly improvements.
Voice Projection Channels
What has us most intrigued is that Einstein works on smartphones, with no need for specialized hardware or commitment to a specific platform. Salesforce seems to be taking its cues from Google in this respect. We have been arguing for sometime now that voice-enablement – regardless of the mode of delivery – is where the money is, especially for software developers. Salesforce is well-versed in hosting a healthy ecosystem of developers, fully aware how they improve the value of the platform upon which they work.
So Salesforce has upped the ante. Alexa and Assistant need partnerships towards building the business-focused installed bases. They might also select for partners who help finesse issues around trust and privacy. Let the games begin.