Einstein Platform: Fuel for AI-enabled World

Brief Overview of Salesforce’s Einstein Platform Services — APIs that allow to build AI-powered apps fast

Recently, Salesforce — one of the world’s progressive CRM platforms introduced a new set of powerful APIs responding to data science communities around the globe.

AI-adopters are exhausted of waiting for stable, versatile enterprise-level solutions to harness the power of the latest AI-based technologies which natively integrated with tools being used across any successful business processes — sales automation, services and marketing.

So, would be nice of Einstein Platform to be one of the most promising candidates to deal with existing machine intelligence integration probs making significant progress towards “AI-For-Everyone & Fast”.

Personalizing Experience Like Never Before — EASY?

No matter if you’re in SMB segment or you represent a part of enterprise-level company — behind each usage of your products stands a real customer expecting relevant experience every new interaction with your business outcomes and processes, and even — harder, better, faster, stronger!


True Positive to Listen: 
Kanye West — Stronger

Let’s Fix that, but — HOW?

Building personalization engines, or at least evaluation of our abilities to do, brings us to one of most common problems which lots of businesses face: how existing data is being historically organized and processed? 
The problem usually appears along with the increasing data volume, total number of data sources, even can be explained as a reaction to positive trends in your customer acquisition and engagement strategies. It’s all OK, but since we want to make data-driven decisions to surround each unique customer by truly personalized environment — all components in our system should process data extremely fast, means to store, enrich, evaluate and score, deliver using different channels according to complex scenarios, re-train classifiers, keep personal levels of coffee incoming to boost as frequently as possible and so on. In addition, we need to take into account that all of these speed-related metrics and factors affect not only on the customer side. Employees also “feel” this effect, as a key executive component in business, since the process in some cases cannot be entirely replaced by systems with desired human-level intelligence. Interesting fact, that decisions recommended by machine have a quite strong level of acceptance, namely:

61% of employees expect artificial intelligence that automates or assists in work-related activities to have a major or moderate impact on their daily work lives. — Salesforce Research


So, we defined a problem related to data processing aspects. Luckily, there’s almost always someone around us who seen the problem earlier and done everything to find and developed an appropriate solution, named DMP 🔥

Data Management Platform (DMP) allows you to collect, organize, and activate your data from any source to gain a holistic view of your consumers in order to power more efficient and effective marketing campaigns and more relative content — source lotame.com

On existing market you can easily find some strong data management solutions depends on your needs, for instance — IBM DM Services, Microsoft Azure, Lotame, of course Salesforce DMP and others.

No one said that this would be easy, but it was important to start from DMP part along with some aspects from personalization point of view, because all this tasty AI-Staff built to deliver insights and as a result — better customer experience, often work together.

Let’s finally move forward to options that Einstein Platform Services brings to the table.

Einstein Platform Services — Brief Overview

by Anna Liednikova

Here we start with definition of what exactly the Einstein Platform Services included, because not all components which bring advanced AI capabilities directly into Salesforce Platform available as separated services by this moment.

Current service portfolio includes:

  • Einstein Vision
    components: Image Classification, Object Detection (pilot).
  • Einstein Sentiment (beta).
  • Einstein Intent (beta).

Einstein Vision

Object Detection is a pretty sophisticated image recognition component, and as for us, the most remarkable and powerful in portfolio in terms of possible use-cases. It brings to developers detailed image attributes detection, which can be used to gain variety of useful insights about your customers, for instance lifestyle preferences (using social media images).

In case you trained it properly, it can count objects, recognize color and size and more. As a possible use-case can be development of inventory apps to solve routine logistics tasks, such as measuring share of shelf in a store or identifying models for customer service calls.

Here is an example how Einstein Vision can be used to simplify products filtering that best match customers preferences.

source: salesforce blog

In case you decide to play with Einstein Vision (we think, you definitely should, after reviewing its API’s documentation), you need to know that usage of prebuilt models allows to save time until you train own models.

Einstein Sentiment

Currently, the component determines the tone of any text (positive, negative, neutral). Potential use-case for any customer support team pipeline can be usage of the component by processing incoming text from emails, ticket-systems as positive, negative or neutral so that team can quickly gain insight into customer attitudes and as a result take actions needed to avoid customer churn etc.

Einstein Sentiment is a result of fundamental outcomes in Natural Language Processing (NLP) research mixed with new types of neural networks. Richard Socher, chief data scientist at Salesforce, says that it’s powerful enough to deal with complex wording and even sarcasm!

We highly recommend you to check Salesforce research blog. Here you can find many interesting papers related to AI and NLP fields. For instance, researcher Bryan McCann explains how contextualized word processing can be created to solve common translation problems.

Einstein Intent

Defines underlying intent of customer inquiries to automatically route leads, escalate service cases and personalized marketing strategies. Follow this Quick Start easily can make you familiar will classifiers creation processes.

Custom classifier creation process that detects intent in text looks like:

Summarizing

Einstein Platform Services, even on this stage, empowers AI-adopters not only with cutting edge solutions — it widely extend our understanding of how the world will be working in the next decades powered by Artificial Intelligence.

source: https://www.salesforce.com/eu/products/einstein/overview/
The important thing is to not stop questioning. 
Curiosity has its own reason for existing. — Albert Einstein