So Salesforce Einstein is pretty much a buzz word recently. And I have completed the related trailhead for it. It seems to me apart from its machine learning and data mining technologies underneath the ground, the functionality are majorly just showing predict result/scoring/likelihood in our current page, and of course, plus sending different emails to different individuals.

So it seems to me that it isn't much to learn about Salesforce Einstein. In the Trailhead, it mentioned:

Developers, you can use the PredictionIO as a Heroku service to build custom machine learning engines in a fraction of the time. Or use the Predictive Vision Service API (beta only) to train deep learning models to recognize and classify images.

But PredictionIO is a service which is not quite associated with Salesforce platform so I am kind of thinking it is not that part of Salesforce Einstein.

My question is, as from a developer or success manager's perspective of view, is there any new feature we need to pay attention to?

2 Answers 2


I have searched a while. Just like lightning experience a few releases ago, Salesforce Einstein has a bright future and will keep on releasing new stuffs in the future releases.

The features which are Generally Available now are on this trailhead: https://trailhead.salesforce.com/en/get_smart_einstein_feat/get_smart_einstein_feat_tour

Personally, I wouldn't consider PredictionIO to be part of Salesforce Einstein unless Salesforce is determined to say so.

To summarize:

Sales Cloud Einstein

Recommended Follow-Ups

  • Get an automatic reminder task to follow back up with customers who haven’t responded to emails.
  • Review the reminder tasks, and accept or reject tasks on demand.

Marketing Cloud Einstein

Predictive Scoring

  • Score every customer’s likelihood to engage with an email, unsubscribe from an email list, or make a web purchase.

  • Help marketers better anticipate the needs of every customer so they deliver the perfect one-to-one journey.

Predictive Audiences

  • Show multiple predicted behaviours in common to marketers when they’re building new audience segments.

  • Create the perfect audience segment to drive customers to the next level of engagement or conversion.

Predictive Content

  • Recommend the best product, content, or offer for each individual.

  • Increase average order values, convert more anonymous web visitors,
    and surface relevant content faster.

Sentiment Insights

  • Automatically analyse the tone and sentiment of conversations in 10+ languages.

  • Make smarter decisions around campaigns and truly understand the voice of your customer.

Language Insights

  • Detect and classify what language a post was authored in, giving marketers the ability to harness all conversations about a particular topic.

  • Slice and dice data into language segments.

Spam Detection Insights

  • Identify known spam sites across the social universe.

Community Cloud Einstein

Recommended Experts, Articles, and Topics

  • Suggested posts, articles, experts, and topic pages.

  • Optimized community experience for each community member, tailored to their interests.

Automated Service Escalation

  • Automatic case creation for customer posts that don’t receive a timely response.

  • Automatic case creation for customer posts that contain specific words (for example, “broken”).

Newsfeed Insights

Analytics Cloud Einstein

Smart Data Discovery

  • Discover insights from millions of data combinations in minutes by automatically examining all possible variable combinations in a data set.

  • Eliminate the manual trial-and-error process of traditional hypothesis-driven analysis.

Commerce Cloud Einstein

Product Recommendations

  • Unique, personalised recommendations throughout the shopper journey, including mobile and desktop e-commerce transactions and in-store interactions with store associates.

Predictive Email

  • Provide tailored content for individual emails.

  • Use relevant content, products, and offers without replacing email service providers.

Commerce Insights

  • Transform vast amounts of product, order, and customer data into actionable insights.

  • Understand product purchase correlation and power smarter merchandising and store planning.

  • I don't see any set date yet as to when Einstein will be generally available. Is Einstein and Salesforce NLP still considered to be in 'beta' at this stage? Commented Jan 18, 2017 at 10:43
  • How do they do automatic sentiment analysis? They will definitely require manually tagged training data for that. Do we as customers need to tag them manually and then use the sentiment analysis API as a service? Commented Mar 30, 2017 at 13:01

Apart from the ones described ,for developers below are the features available which should be treated as a part of Einstein for developers

PredictionIO in Heroku Private Spaces

Although PredicitionIO is an open source framework ,Salesforce’s Heroku Enterprise is now made compatible to support this .This will allows developers to build custom intelligent applications using an industry leading, open source machine learning framework

There is an example from the Salesforce Evangelist team to show how one can achieve this.

Predictive Vision Services and Predictive Sentiment Services

There are API available on this as a part of Pilot program and nothing in GA.You can explore the documentation here

We will need to watch out more in coming releases .For now the information thats available online is very thin

Predictive Modeling Services

Apache Kafka on Heroku

Kafka is generally available now as a service on Heroku platform and it supports Big data event handling .With Kafka, developers now have the ability to work with streams of billions of events in real time helping make their apps even smarter.

Prediction.IO was acquired by salesforce and they will provision wrapper around the solution to make it easier for developers .Think of it as how java and apex have similarity , aura and lightning and in same way Prediction.IO and Einstein .

Again all these are getting started and as time goes along and SFDC like any other new technology will invest into their R&D and hopefully we will see these getting matured with every release.

I have derived most of the content from here and this neat blogpost

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .