Salesforce, which has been dribbling out details of its grand plan to add artificial intelligence to its sales-and-marketing software for weeks, did a bigger reveal Sunday, a few weeks before its Dreamforce event and—not coincidentally—just as rival Oracle kicked off its annual Oracle OpenWorld tech conference. The timing is probably not a coincidence.
is adding more intelligence to each of its branded Marketing, Sales, Service, Commerce, and App clouds, with a mix of AI technologies built in-house and acquired in a series of acquisitions, including of RelateIQ in 2014 and Metamind last April. Salesforce Einstein, as the effort is known, seeks to help Salesforce customers wring real value out of the tons of data they generate.
Is there a major client that you haven’t heard from in awhile or who has been name-dropping your competitor in email or or on social media? It’s time to reach out. Do you want to see if your branding is getting out there? Why not pull every mention or relevant photo posted on Facebook or Twitter
or other social network? That’s some of what what baked-in AI makes possible.
Get Data Sheet, Fortune’s technology newsletter.
If you have hundreds or thousands of sales prospects on a list, how do you tell the potential winners from the duds? Salesforce says its Marketing Cloud’s Predictive Scoring will prioritize which prospects are most likely to actually write a check over the tire kickers.
“Every sales person wants to sell more, they want more qualified leads, not window shoppers,” Richard Socher, chief scientist at Salesforce, told Fortune. “We combine CRM information with e-mail and calendar data. We can detect if a competitor is mentioned on an email thread.”
Not only that, the software can do a better job tracking responses (or lack thereof) to marketing pitches, he said. “Machine learning will flag us if a C-level person is not responding.”
Salesforce started down this path a few years ago and is using a combination of in-house and acquired expertise to build AI for CRM, according to John Ball[, senior vice president and general manager of Salesforce Einstein.
Even before the flurry of acquisitions, Salesforce was working on some of its own AI projects and fielded a multi-tenant platform capable of running this technology. But Salesforce also identified specific technologies for acquisition, he added. On Salesforce’s most recent earnings call chief executive Marc Benioff boasted of the team of the 175 data scientists the company now has on staff.
AI, and the whole idea of helping companies derive actual financial benefits from their data, is a big push for nearly every tech company. Amazon
(with its ubiquitous Watson), Microsoft
are all pouring resources into making their own software smarter and able to automate tedious tasks formerly performed by humans.
Salesforce says it can apply AI technologies like machine learning, to sort through an ocean of available data —from email, chat, social networks, traditional databases—and prioritize it for its customers, in this case sales people, marketers, and service personnel.
For more on Salesforce read 5 Things to Know from Salesforce earnings call.
“We have tons of data on our accounts, opportunities, communications around email and calendar and interaction data like campaign emails and social data from Social Studio, Community Cloud and Chatter,” said Socher, who was formerly chief executive of MetaMind. “Einstein uses deep learning, machine learning and natural language processing to those rich signals. With that we can surface predictions in the context of the business user.”
Salesforce Community Cloud lets partners and customers share information. Chatter is a Slack-like messaging service for employees, and Social Studio lets customers keep an eye on social media commentary about their companies.
For more on Marc Benioff watch:
Each piece does its part. RelateIQ, for example, continuously evaluates incoming data about salesperson-client interaction and, if it notes a lack of communication/feedback in a set amount of time, recommends a call or other outreach. Sources said after Salesforce bought this company two years ago, Benioff wanted to affix “IQ” to each of the company’s clouds, but the advent of IBM Watson and the marketing hype it’s generated, caused him to switch to Einstein. So what would hav been the MarketingIQ cloud is now the Marketing Einstein cloud.
Meanwhile, MetaMind, combines natural language processing, computer vision, and database predictions in a way that will let marketers search for data relative to any topic. If that topic happens to be soccer, it can search across social media not only based on language and relevant company names, but on images. That way it can distinguish between posts about soccer (aka football in most of the world) and American football since the software can differentiate between a round soccer ball, and the more elliptical American football.
The new AI capabilities will run both in Salesforce’s own data centers and on its products that also run on Amazon Web Services, Ball said. Last May, Salesforce said it inked a deal to use AWS infrastructure to run a bunch of its new software.
A lot of Einstein roll-out details are still to come. Some of Salesforce’s new AI features will be blended into existing cloud offerings at no additional charge. Others will carry an additional cost. Many of the promised capabilities won’t launch till next year.
The predictive scoring for marketers as described above will be available in October, as will a “predictive audience” feature that claims to forecast likely behavior of a given audience segment based on current data. Both will be part of the marketing cloud. Also available in October will be a product recommendation feature for the Commerce Cloud that will tailor product recommendations for shoppers based on their mobile, desktop, and in-store interactions. Salesforce Commerce Cloud is basically Deamndware, the e-commerce company Salesforce bought for $2.8 billion in June,
Both Socher and Ball repeatedly noted that AI from all these sources will be tightly integrated into the company’s core clouds soon. Still, given that many of the relevant companies—including Demandware, MetaMind, Implisit, and TempoAI—were acquired within the last year, it’s hard to see just how deep that integration can be. Customers will be able to judge for themselves next year when more of the capabilities come online.