Autonomous Insights

Zuye Zheng 14. September 2020 Add Intelligence, Winter 21 0

Since the launch of Einstein Analytics over 5 years ago, we’ve been working hard to make the authoring experience richer and more declarative with “clicks, not code”. We hope this has empowered more of you to become dashboard builders and helped you build powerful analytics apps faster for your entire organization.

We’ll be taking the next step in this journey to democratize with Autonomous Insights – to bring your entire organization into this data culture with self-service analytics augmented by machine learning (ML) that is integrated and learning from your business.

Note: Autonomous Insights is now known as Ask Data for Salesforce.

How AI can jump start self service analytics?

Einstein Analytics has been focused on providing the best analytics experience on top of your Salesforce CRM data and we are continuing this focus with Autonomous Insights. With the Salesforce platform, where users across your business login to the same instance to collaborate on shared assets and data, we have unique challenges for analytics, but also interesting opportunities to help alleviate the burdens of IT setup and admin curation when coupled with machine learning.

The first, and often biggest, challenge to providing analytics in a shared environment is just finding the right data to get started. Complex shared schemas are compounded by a growing set of different versions of the right data in the form of Report Types and Datasets. It’s this “getting started” problem that we’re tackling first with Autonomous Insights. We want to get you to the right data for actionable insights as fast as possible, regardless of whether you’re a casual user or an analytics pro, and we want to ensure that you can ask questions about all your data from the start.

This might sound pretty hard with some Salesforce users accruing years or even a decade worth of assets. However, it is this very fact that we will leverage to help solve this problem.

Einstein Analytics is not a standalone analytics product, it’s native to CRM and part of the suite of Salesforce products you use to run your business. It’s where all of your users have worked together to build, refine, and tailor to your specific business needs. With this, you have already told us the data that is important to you and the language you use to talk about that data with your coworkers. We will be using all of this information to build a self-service and autonomous analytics solution just for your business.

A search tailored for analytics

Before we dig a little deeper into how it all works, here’s a quick preview of what it is. It all starts with a new, search-based insights experience that will suggest user-specific recommendations – influenced by not only the users’ interactions, but by what’s going on across the organization – to help you pick up where you left off, explore new insights, and find new collaborators.

Screenshot from 2020-09-01 19-27-17.png

From here, you can search all of your data for insights, using your own language. We take a flexible approach to language understanding and use models trained with the language of your organization found across Salesforce so they will be tailored to how you talk about your data with no initial setup.

From a single search, we’ll generate insights in the form of a natural language query, provide further drill-downs using our next best query models, and find existing insights with both a Salesforce search and our new analytics semantic search powered by our semantic graph.

Screenshot from 2020-09-01 19-29-40.png

A semantic graph to make sense of everything

Before we can leverage all of the information and usage across a Salesforce instance, we first need to connect the dots. Our new semantic graph maps how data is interconnected and how users use it across different Salesforce silos, from Salesforce Objects to Classic Reports to Einstein Analytics.

The semantic graph allows us to provide user-specific, but also community-influenced, recommendations; a semantic search that considers data lineage, not just how things are named; and the ability to apply transfer learning so that you can interact with all your data in the language you’ve already been using in Salesforce.

To help visualize one such application, below is the nebulous semantic graph (left) of one of our internal instances. We’re able to apply graph modularity (right) to form clusters of semantically related datasets, assets, and users to provide better recommendations as well as understand which of the hundreds or thousands of fields in a dataset make it the most relevant for a particular question.

Screenshot from 2020-09-09 10-08-16.png

Learning and the feedback loop

We believe a big part of removing the barrier to entry to analytics is through natural language interfaces and specifically allowing you to use the language of your business. With all the interactions you’ve already made with all of Salesforce, not just analytics, we already have a good idea of how you talk about Opportunities vs.Opportunities with Products vs. the custom objects and fields that make Salesforce such a customizable platform. With our semantic graph and neural networks, we piece all of this together to build language customized for each business without any upfront IT setup.

Screenshot from 2020-09-08 19-51-46.png

Leveraging all of the interactions with Salesforce, we get a good head start, but no autonomous system is perfect. However, great autonomous systems can adapt to learn from the collective knowledge of an entire organization.

This is why a feedback loop and relearning is fundamental to Autonomous Insights, we want to ensure no interaction is wasted and that each goes into improving the system. Our user experience is designed around easy refinements and corrections to the natural language understanding and recommendations which will all go back to improving and refining your models. This allows us to crowdsource improvements from across the organization, which also unburdens the handful of admins and experts from defining everything.

Screenshot from 2020-09-02 14-03-52.png

Enterprise data governance

Mapping and learning from all your interactions with Salesforce can start sounding a little scary, but the same data governance you’ve come to expect with Salesforce has been engineered in from the start. Salesforce and Einstein Analytics sharing rules are fully respected and all models are built specifically for each org, and only that org, with the data never leaving our secure data centers.

Looking ahead

This is just the first step for us and we look forward to working with you to refine the experience and product so that you and your business can get the most out of self service and Autonomous Insights.

*Forward-looking statement

This content contains forward-looking statements that involve risks, uncertainties, and assumptions. If any such uncertainties materialize or if any of the assumptions proved incorrect, the results of, inc. could differ materially from the results expressed or implied by the forward-looking statements we make.

Any unreleased services or features referenced in this document or other presentations, press releases or public statements are not currently available and may not be delivered on time or at all. Customers who purchase our services should make the purchase decisions based upon features that are currently available., inc. assumes no obligation and does not intend to update these forward-looking statements.

How useful was this post?

Click on a star to rate useful the post is!

Written by

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.