Artificial intelligence is already all around us, but is more often used by consumers than by big business. Think Microsoft's Cortana, Google's Google Now, and Apple's Siri. These tools rely on language recognition and a dash of smarts to help people complete simple tasks (send messages, set calendar appointments) on their smartphone, tablet, or PC. Salesforce believes it is time to bring artificial intelligence to business apps on a grand scale, and promises that it will be easy to use.
"For most business people, AI has been too complex and out of reach," said Salesforce General Manager John Ball during a call with media. "You have to collect and integrate a lot of data, convert it to a specific machine format and hire scarce data scientists to work on it and have an infrastructure that's secure and scalable. Even if you have all that, the last mile where a lot of AI projects get tripped up is you have to be able to surface the insights in the context of your business applications -- that's just too hard for the vast majority of companies out there."
Salesforce is adding its Einstein artificial intelligence product to the bulk of its enterprise software offerings. The move, says Salesforce, is meant to help people build AI-assisted applications that rely on the smarts generated by machine learning, predictive analytics, and natural langauge processing. The end goal will be what Salesforce says is the "world's smartest CRM" that allows every Salesforce user to deliver the best possible expereince to his or her customers.
Einstein is deeply embedded into the Salesforce Platform. App Cloud Einstein is the core tool that will allow companies and developers to add artificial intelligence to their own apps through existing Salesforce developer tools. The primary targets are data scientists and developers who can use Einstein to soak in and analyze customer data. Einstein's AI powers will be able to plumb through every Salesforce cloud application and process any form of data, including customer details, communications, social networking interactions, and eve IoT device data.
Some of the key functions will be predictive vision services, predictive sentiment services, predictive modeling services, and PredictionIO in Heroku private spaces. In predictive vision services, for examples, developers will be able to train software to recognize and classify images. Developers can further train the software to monitor and react to large volumes of images at scale that humans simply cannot. For example, a data scientist could create a model to recognize different car brands on Facebook or Twitter in order to identify brand advocates.
Salesforce says Salesforce users won't need to know code.
"Business users and admins will be able to embed AI fields to customize any existing Lightning apps, and build entirely new AI-powered apps with clicks, not code," explained a Salesforce spokesperson to ProbrammableWeb via email. "For example, a predictive score field can be added to any page, process or rule, while end users can build reports or list views sorted by score."
Developers will be able to leverage the API to mine Salesforce for almost any type of data and turn it into intelligence. The company was very short on details of if/how the API and SDK are changing. Those are expected to surface during the company's Dreamforce conference next month. Until then, we know the AI features will be added to Sales Cloud Einstein, Marketing Cloud Einstein, Commerce Cloud Einstein, Service Cloud Einstein and Community Cloud Einstein. The company will provide in-depth previews of these apps at Dreamforce.