Dreamforce 2017
As every year comes to Winter, it comes Dreamforce, the best place to be to gather a glimpse of what’s going on in Customer Success platforms. The first day opened with almost 190K attendants joining Salesforce team in San Francisco to learn the upcoming changes that this amazing platform is going to bring to Digital Customer Experience.
Einstein Platform Services Overview
Showed some use cases that Einstein Platform can provide out-of-the-box. They adopt an intermediate approach between out-of-the-box services such as Google Vision API or AWS Rekognition and low-level frameworks such as Google TensorFlow, Theano or AWS GlueOn. Salesforce offers a set of pre-trained models directly usable inside our applications to complete specific tasks (food recognition, animals, customer complaints, etc.)
Using Apache Prediction IO to predict students drops at University
This is a great use case for Salesforce to show their Apache PredcitionIO (https://predictionio.apache.org) which they’ve acquired and contributed last year to the Apache foundation. In particular, it shows how a predictive model can be deployed to PredictionIO to build and train a predictor. Salesforce offers a fast deployment path through Heroku. However, PredictionIO is a Scala Framework with a set of machine learning algorithms that can be trained. The speaker demonstrated how a Naive Bayes algorithm could predict students drops with an accuracy of 81% using only a small set of 150 records. Moving on, he uses sci-kit-learn (http://scikit-learn.org/stable/) framework to perform dimensional reduction and publish the resultant code as a REST API through Apache PredictionIO Service.
Salesforce Metadata
A completely redesigned support for Object Metadata, enabling creation and management of either Salesforce or custom objects within Apex code. This opens capabilities to a new set of applications that could extend Salesforce Platform (or Cloud as it is called these days) creating infrastructural extensions (think about a Heroku Connect component but mapping to other services.
Marc Benioff on Stage — Dreamforce Keynote
It started as usual with the traditional Aloha dance, Marc please change this, we cannot afford one more year of it. Then Marc Benioff hit the stage showing how we’re into the Fourth Industrial Revolution which makes a pivotal point on Intelligence, where companies are disrupting environments such as tire production or soda coolers with IoT and Computer Vision. Has never been more exciting being a developer than now.
Listening keynote made clear Marc Benioff is going to be President one day.
However, Benioff is a great speaker and, together with Parker Harris, showed enhancement of Salesforce Cloud:
Salesforce is going to release an upcoming update to its architecture, making this the very first Event based application platform. This makes their cloud platform riding the new trend of event-based stream technologies.
Einstein prediction builder means people without any Machine Learning knowledge can customize prediction models selecting attributes within your data and re-training the ML model behind the Sales Cloud to predict acquisition, attrition or churn rates.
MySalesforce means salesforce took a huge step back in company branding, allowing customers to deeply customize platform with their brand identity and even publish a custom branded app.
MyTrailhead represents the next generation employee learning platform, allowing users to deploy custom training content into a company-branded Trailhead environment
It has been a great first day. Can’t wait to tell you what’s coming next. Stay tuned for Day 2!
