Connections

NEW YORK — Salesforce unveiled new Einstein Platform Services that it said at the second annual O’Reilly Artificial Intelligence (AI) conference June 28 will enable developers to bring natural language processing (NLP) and image recognition to any app, creating AI-powered customer experiences for sales, service, marketing and commerce.

Developers can use Einstein Platform Services to quickly build custom AI-powered customer relationship management (CRM) apps, according to the company. The additions to Salesforce’s Einstein Platform Services that were launched June 28 will enable developers to create custom deep learning models to fit their specific business needs, Salesforce chief scientist Richard Socher said during a conference keynote.

One of the additions is Einstein Sentiment, which Socher said enables the creation of smarter apps with NLP. Users can leverage pre-trained models to classify text as positive, negative or neutral, and Sentiment seamlessly integrates with Salesforce workflows, allowing users to better understand their customers, according to the company.

Sentiment will allow developers to “classify the tone of any text—from prospect emails and social media posts to customer reviews and message boards—as positive, negative or neutral so that companies can quickly gain insight into customer attitudes and then take appropriate actions,” Salesforce said in a news release.

For example, an airline can create a custom app using Sentiment to analyze tweets and other social media to understand overall impressions and prioritize negative posts that require service responses, it said.

Einstein Image Classification and Object Detection, meanwhile, allows users to access pre-trained deep learning models or train their own to recognize and classify images, Salesforce said.

Object Detection enables developers to train models to recognize multiple unique objects within a single image, as well as the location, size and quantities of those objects, according to the company. Object Detection can be applied to improve service, inventory management and retail share-of-shelf apps, it said.

For example, a beverage company can streamline inventory management and service by automatically analyzing photos of retail shelves to count the product, and then calculate and process a new order, it said.

Another new addition to Einstein Platform Services is Einstein Intent, which Salesforce said in its news release will allow developers to train a model to classify the underlying intent of customer inquiries to automatically route leads, escalate service cases and personalize marketing campaigns. For example, a retail company could use Einstein Intent to build a custom app that automatically classifies inbound customer support queries to identify which customers are experiencing shipping problems and then proactively provide support messaging and tracking details, Salesforce said.

In a separate session at the conference, Socher said his company was still hard at work tackling the “limits” of single task NLP and deep learning. For one thing, he said, despite the “amazing improvements” that have been made in AI and deep learning, “if we continue to start all our projects from some random parameter setting, we won’t ever obtain a single model that captures all the complexity of natural language.”