Salesforce chief Marc Benioff isn't very good at keeping secrets. Or more realistically, he intentionally leaks bits to the press to get everyone excited about the news he'll be making at Dreamforce, his company's massive user conference that takes over San Francisco each fall.

Two years ago, the conference team draped a banner across the entrance to the Moscone Center convention complex that identified Salesforce as "The Customer Success Platform" in advance of Dreamforce — then acted surprised, no, horrified, when the press started writing about its alleged strategic shift. 

This week we found out Salesforce will be unveiling a new Artificial Intelligence (AI) product called Salesforce Einstein at this year's Dreamforce. The news comes via a Forbes Magazine story by reporter Alex Conrad — where Benioff muses that Salesforce Einstein is poised to be "the next big thing" — and from a related tweet from Benioff himself.

What We Know About Salesforce Einstein

Benioff, being the showman that he is, was careful to give away just enough details to pique everyone's interest — and not enough to satisfy anyone's curiosity. So now we're forced to do exactly what he wants us all to do: to ponder and debate and generate even more interest in what he'll reveal at Dreamforce. 

At least we understand the game plan. 

With that in mind, it's clear is that Einstein won't be yet another Salesforce cloud. Instead it will add business intelligence to the other clouds by making it possible to direct salespeople toward the right customers with the right products and the right deals at exact the moment they are most likely to buy.

Is that really all that revolutionary? Hard to say.

Constellation Research analyst R. Ray Wang, in an article we wrote about Salesforce's BeyondCore acquisition last week, told us Benioff likes to raise his Total Addressable Market size (TAM) before Dreamforce. "Last year was (the Internet of Things) IoT, the year before it was big data and analytics. The year before, customer experience and customer success. This year, it's AI for him. These series of acquisitions are to have him catch up on AI," he said.

The acquisitions in question include RelateIQ, which combines CRM with data science, smart calendar app Tempo AI, MinHash, which offers assistive intelligence and data science for enterprises, PredictionIO for machine learning and MetaMind for deep learning.

Salesforce Needed AWS

But software only goes so far. To crunch big data in real time and to apply machine learning to get AI, you need massive compute or a powerful cloud. That’s why Salesforce expanded its strategic relationship with Amazon earlier this summer and made Amazon Web Services (AWS) the company's preferred public cloud infrastructure provider, according to Constellation Research analyst Holger Mueller.

"Salesforce cannot run any kind of modern machine learning architecture in house on its Force.com infrastructure. It needs AWS for that," he said. "Or a completely different new platform architecture"

AI is more complicated than software and cloud, Wang explained, noting that there are seven components: massive compute power, natural UI/UX, time, industry specific expertise, large corpus of recommendation engines and awesome math talent.

IBM, OpenText, SugarCRM, Microsoft ... One Big AI Party

Benioff is hardly the only tech executive who thinks AI will change the world. There's IBM CEO Ginni Rometty with Watson, OpenText CEO Mark Barrenechea with OpenText Magellan, SugarCRM CEO Larry Augustin with SugarCRM Intelligent Agent Candace, Microsoft CEO Satya Nadella with Bing-powered Cortana ...

But as Constellation Research analyst Alan Lepofsky reminded CMSWire, "Conference season is only beginning." 

After VMworld next week, there will be Teradata Partners, BoxWorks, Microsoft Ignite, Oracle Open World, Dreamforce and SAP Tech Ed. Expect AI to be a hot topic.

And there's something far more important, beyond AI, so don't get bound up in the technology, warned Lepofsky.

"To turn the hype of artificial intelligence into reality we need to focus on explaining what it means to employees using the software. The only people that care about the details of the technology are those making a vendor selection and perhaps developers choosing a platform," he said. 

"For end users, the benefit is to simplify the way they get their jobs done. AI can help with things like automating repetitive or mundane tasks and filtering information to provide only what they need to see in context of what they're doing."

Needless to say, we'll hear a lot more.