Salesforce debuted a range of new capabilities this week geared at helping developers build custom artificial intelligence solutions for their enterprise clients.
The CRM leader's Einstein platform has seen several upgrades since its introduction two years ago—mostly out-of-the-box capabilities to enhance its comprehensive Software-as-a-Service portfolio.
While those releases embedded various AI tools into Salesforce's Sales, Service and Marketing clouds, enterprises increasingly need to build their own applications that leverage advanced machine learning to tackle complex business challenges.
That's why "we're now giving increasing love to Salesforce developers with Einstein Platform Services," Marco Casalaina, vice president of product for Einstein, told CRN.
The industry's shortage of data scientists has challenged enterprises looking to solve unique business problems with AI.
"That's the problem we're looking to solve," Casalaina said.
"We can counter this massive shortage of data scientists by providing these capabilities directly in Salesforce in a way that's easier to use than traditional custom AI and custom machine-learning toolkits"
The first additions to Einstein Platform Services are translation and OCR (Optical Character Recognition) features that can be accessed by developers, especially those using Apex, the native Salesforce programming language. Other services will follow.
The data custom applications generate using those features never leaves the "trusted Salesforce environment," Casalaina said.
Salesforce is also launching Einstein Predictions Services—a toolkit that makes it possible for applications from other vendors to access Einstein to make predictions.
"A new API allows these predictions to be calculated and used in real time in third-party platforms," Casalaina told CRN.
Salesforce has also developed a technology that imparts "trusted AI," he said, empowering customers to monitor and remove unwanted bias.
"Ethics in AI is a thing these days," he said.
Einstein has always been transparent about how it arrives at its predictions. But eliminating bias, especially second-order bias, isn't as easy as it sounds, Casalaina said.
That's why Salesforce has introduced the notion of "protected fields," he said.
Designating certain fields as such allows customers to not only exclude certain data from calculations, but also to eliminate any results that correlate to that data.
With the new Einstein capabilities, Salesforce partners are expected to start developing and integrating more powerful solutions that deliver to their customers custom predictive services, he said.