Dreamforce 2017 Shows Progress in Salesforce’s Push into AI

Company Steps Up Investment in Einstein Platform

A few weeks ago, I had the pleasure of attending Salesforce’s annual Dreamforce event in San Francisco. The event saw an astonishing 170,000 attendees, made up of customers, partners and members of the Salesforce community.

Salesforce is firing on all cylinders at the moment following a monster quarter in which revenue grew 25 percent, making it the fastest enterprise software company to reach $10 billion in annual revenue.

One of the most interesting areas of the future for Salesforce is its artificial intelligence platform, Einstein, which celebrated its first birthday at Dreamforce this year.

Einstein Has Had a Busy 2017

Salesforce has spent the best part of the past year working hard to bake Einstein more deeply into its core customer relationship management (CRM) platform. In total, it shipped 20 artificial intelligence features in 2017, 16 of them new, aimed at sales, marketing and customer service. They include applications such as engagement scoring for marketing, service bots and intelligent sorting for field and customer services, and sales forecasting and predictive lead scoring for sales representatives.

Additionally, for developers, Salesforce launched a set of new tools for building custom predictive models using artificial intelligence. It also added new vision and speech APIs for apps built on the Salesforce platform.

Here’s a snapshot of the impressive feature enhancements in Einstein shown at the event.

Click on the image above for a larger version.

Customer Adoption Is Growing

Salesforce also gave a few updates on the adoption of Einstein. The platform is now producing more than 475 million predictions a day. Customers using Einstein include Adidas, AXA Insurance, US Bank and Weight Watchers.

Click on the image above for a larger version.

The biggest use of Einstein is providing intelligence for sales, especially forecasting, lead scoring and account insights. Salesforce boasted that its efforts over the past year to make Einstein easy to use was helping “turn sales people into data scientists”.

US Bank’s head of CRM, Bill Hoffman, joined John Ball, general manager of Einstein, on stage to show how Einstein is improving sales closure rates by 300 percent compared with its previous scoring methods.

Another major Einstein customer is Adidas, which used the Dreamforce event to launch a new version of its shopping app, which draws on Einstein to provide a more personalised customer experience. Adidas will use machine learning to help improve customer interactions by collecting data on preferences and making recommendations to customers to buy new products. Adidas will also use Einstein to sort and prioritize customer complaints and plans to launch a chat bot for customer inquiries within the app.


One of the central themes in CEO Marc Benioff’s main presentation was Salesforce’s strategy to enhance customisation and personalisation within its platforms. This approach is vital in the context of Einstein and artificial intelligence as well. Demand is high to build custom artificial intelligence solutions in businesses at the moment, but skills are in short supply and expensive. The average entry-level salary for a junior data scientist today is more than $100,000. An experienced worker in this field can command more than $200,000.

The big Einstein announcement at Dreamforce 2017 therefore was the launch of myEinstein, a set of services that enable Salesforce administrators and developers of all levels to build custom and personalised apps with artificial intelligence on the Salesforce platform.

Salesforce also launched Einstein Prediction Builder, which allows customers to automatically create custom models that can predict outcomes for custom fields and objects in Salesforce. It announced new features for image recognition, language recognition and discovery, as well as set of bot services that can be trained to improve customer service workflows by automating tasks such as answering questions and retrieving information.

These are important steps. We have long argued that artificial intelligence tools for enterprise developers need to come down from the technical and academic ivory towers and into the real world. Salesforce is smart to lower the technical barriers to entry and make tools more transparent and simple to use.

Salesforce’s principal users are sales professionals who need to build and operate artificial intelligence “with clicks, not code”. US Bank showed how it set up Einstein’s predictive lead-scoring features, used for 4.5 million leads, in under two hours as an example.

What Does 2018 Have in Store for Einstein?

The Dreamforce event showed evidence of good progress in artificial intelligence, and Salesforce has spent much of the past 12 months building features and educating customers.

Artificial intelligence is still in its infancy within enterprises, the market is driven by proofs of concept and experimentation at the moment. This is reflected by the huge activity around the topic among the open-source community in GitHub, for example.

We would expect with these foundations in place 2018 will reveal more adoption of Einstein, particularly in sales forecasting and lead scoring.

A revealing section of the event’s presentations came from Richard Socher, chief scientist at Salesforce, who showcased some the firm’s artificial intelligence activities that will become major focus areas in 2018. Mr Socher, previously the CEO of MetaMind which was acquired by Salesforce in 2016, discussed efforts to improve speech processing and translation. He demonstrated improvements to Salesforce’s text summarization and image processing technologies, which he claimed are now among “the best in the world”.

A future opportunity will be sentiment artificial intelligence applied to CRM. Salesforce how its intent analysis is playing a part in predictive lead scoring. Sentiment analysis is already playing a role in Salesforce’s offerings, including its Community Cloud. We believe that, coupled with its intent API, sentiment insight will be the next frontier for Einstein’s sales and marketing customers, and could be a big source of advantage for the company in the future.

Google Partnership Will Be Crucial

Lastly, as the most important strategic announcement at Dreamforce 2017, it will be fascinating to watch how the partnership with Google Cloud will evolve in terms of developing artificial intelligence. For now, Google and Salesforce are confining the partnership to infrastructure as a service, CRM, G Suite and Google Analytics.

But we expect further collaboration and ultimately a deeper integration of Google’s artificial intelligence tools with Einstein. This become more likely as the Einstein platform continues to open up to developers and Salesforce gains more experience in working with others on artificial intelligence, thanks to its relationship with IBM’s Watson. A union with Google could form a formidable challenge to Microsoft in this area.

In summary, Dreamforce 2017 showed that Salesforce, like Amazon, IBM, Google and Microsoft, is definitely one to watch in the nascent market for enterprise artificial intelligence.