AI is a class of technology that teaches itself how to perform tasks. In one form, AI can learn to act as a personal assistant to its human users. But to date, much of the available AI technology has been expensive and hard to implement.

At the company’s annual Dreamforce event in San Francisco, Salesforce (CRM) chief executive Marc Benioff and other execs again talked up the company’s plan to integrate AI into its various sales, marketing, e-commerce, and customer service software products.

And some customers at the show said they were eager to try out the new AI perks, lumped under the “Einstein” label. That’s despite the fact that it’s not quite clear which of these smart features will be folded into existing Salesforce software subscriptions at no extra charge, and which will cost more money.

“If a customer buys $500 worth of wine online, I need to be in touch with that person,” said Paul Leary, chief executive of Bespoke Collection, a Napa, Calif. vineyard and wine merchant who uses Salesforce’s Sales Cloud software. Being able to track customers across wholesale, retail, online, and brick-and-mortar store channels would be extremely helpful, he said.

Get Data Sheet, Fortune’s daily technology newsletter.

Patrik Frisk, chief executive of Aldo, a Montreal-based shoe wholesaler and retailer, is similarly intrigued. Aldo, with 13,000 employees and 2,500 stores in 100 countries, uses five Salesforce products, and thinks Einstein could help across the board.

“We have historical transaction data. Now we can look at that through Salesforce by geography—what sells through our store in Burlington, Vermont? What sold via e-commerce there and what sold through our wholesale partners there?” he explained.

All of that data, along with a survey of what local people are saying about the brand on social media and what they are searching for online, would be invaluable in planning what inventory to stock in the region, for example.

Sales Cloud Einstein promises to look at sales leads through the prism of email and social media messages and other sources to rank which are most likely to result in a sale and to recommend immediate follow-up action by the account executive. Less promising leads might be routed to inside sales.

“I don’t tell Einstein about what my prospects are, it just figures them out,” said Shubha Nabar, Salesforce’s director of data science, during the Wednesday keynote. The software will know, based on what it’s seen of a prospect’s history, that this person works in this industry, in this geography, and has viewed a product demo online. It can then even “write” the email message to be sent to this prospect.

Salesforce is promising big stuff here. It claims it’s been able to integrate a variety of homegrown and acquired technologies into its sales, marketing, and service software in a way that everyday people—not just data scientists or other specialized tech geeks—are able to use it.

IBM (IBM), which has been touting its Watson AI brand for years, has poured resources into building businesses around it—in healthcare, in retail, and in banking, but Watson is expensive and hard to implement, said Larry Carvalho, an analyst at research firm IDC. It’s not clear if and when IBM will make money from it.

Ease of use, across the board is a big push at Salesforce as with other business software companies. There was not much news at the Wednesday keynote but the company did mentiona new Live Message capability for Service Cloud that would let service reps send texts via their messaging app of choice automatically as demonstrated in this video.

The aforementioned companies, Aldo and Bespoke, are not only Salesforce customers, they are customers who were willing to spend the time and money to come to San Francisco this week. If they’re not true believers, who is?

To get a better idea of whether Salesforce succeeded here will take more time, as customers try out these features in their own shops. While some of these AI perks are available now, the bulk won’t be out till next year.

For more on Salesforce, watch Fortune’s video:

Some Salesforce customers—not necessarily at the show this week— privately grouse that Salesforce prices are too high. One third-party consultant who specializes in sales software, said he has seen several competitive deals go to Microsoft (MSFT) over Salesforce in recent months, in part due to pricing.

So as is always the case, claims of better, faster, smarter software are standard operating procedure. It’s the execution part that’s harder.