It's been just over a year since Salesforce introduced Einstein, a set of artificial intelligence technologies that are designed to underlie and enhance the Salesforce product set. Today, at Dreamforce, the company's enormous customer conference taking place this week in San Francisco, it announced myEinstein, a package of tools it created to help developers and Salesforce admins customize their AI tools for their particular business.
This is really an extension of the developer tools the company introduced in June this year with the goal of exposing some of those artificial intelligence capabilities to customers and letting them create their own intelligent apps. But Salesforce is not stopping with developers in this release. They are also giving customization capabilities to admins with a forms-based approach that doesn't require any programming skills whatsoever.
John Ball, GM of Einstein at Salesforce, said that this new package of tools allows companies to customize aspects of Einstein in the same way they are used to doing it with other parts of the Salesforce toolset. When a company buys Salesforce, they rarely use it as it's presented out of the box. Instead, they alter it to match their company jargon and workflows.
Sometimes that involves developing applications on top of the platform to extend its capabilities, and sometimes, it's a matter of giving admins a set of tools to reconfigure parts the product. myEinstein includes both kinds of tools.
Salesforce has talked frequently about wanting to put AI in the hands of customers who lack data science or model building skills. They are trying to extend that capability with today's announcement by offering admins the ability to build prediction-based fields with a few clicks. I know the "few clicks" part sounds like a lot of marketing mumbo-jumbo, but if it works as in the demo, it is really just a few forms where you select the fields or objects for which you want to create a prediction and you're good to go.
For instance, you might want to predict which invoices are most likely to go over 90 days or which customers are most likely to buy this month. While it doesn't require data science, it does require at least a partial understanding of AI in order to ask the right questions. It will be interesting to see how far customers can get on their own with this tool without some additional training or help from others inside the organization.
The other new tool enables admins to build, train and deploy their own customer service bots, which use natural language processing (NLP) to understand the customer inquiry and answer simple questions. For example, they could create a bot to gather enough information and direct the customer to the correct customer service person, or even answer questions without getting directed to a human.
While Salesforce again claims an admin can build these bots with a few clicks, there is a danger here because these tools are being used with customers instead of just internal users, and if thing go awry, they do so with customers.
But Salesforce believes they are actually minimizing that risk by greatly reducing the complexity associated with creating bots, while using underlying intelligence to understand customers based on information they know about them stored in Salesforce.
Ball says admins work with these types of customization forms on a daily basis, and giving them access to AI tools is going to offer them a new set of capabilities to enhance the product even more. This is potentially putting a great deal of power in their hands and it will take some real-world scenarios to see how this plays out in practice.
It's worth noting that while the admin tools are being announced today, they are currently being piloted , and won't be available until next summer.
This article originally appeared on TechCrunch.