myEinstein

• This year’s Salesforce Dreamforce conference started rather big with the announcement of my
• Einstein. The news spread quickly through the massive crowd of 170,000 plus attendees at Dreamforce 2017 and was met with equal levels of enthusiasm and scepticism.

The Einstein machine learning platform has been updated with new predictive insights and chatbot capabilities. In ways that could truly make AI and deep learning more accessible to developers.

‘Point, Click, Predict.’

• The latest iteration, Salesforce my
• Einstein, allows users of all skill levels to now develop custom AI apps “with clicks, without being a data scientist.”

The tool has two new services: Einstein Prediction Builder and Einstein Bots. Einstein Prediction Builder enables automatic creation of custom AI models that can forecast outcomes for any field or object in Salesforce. Whereas, with Einstein Bots developers and admins can use a point-and-click interface to build custom chatbots. It is a service which can be trained to augment customer service workflows by automating tasks such as answering questions and retrieving information.

Putting coders out of business?

So how does myEinstein works with ‘simple clicks’ after all? The declarative setup guide walks users through building, training and deploying AI models using structured and unstructured Salesforce data. The service automates the model building and data scoring process and custom predictive models and bots can then be easily embedded directly into Salesforce workflows. Models and bots automatically learn and improve as they’re used, delivering accurate, personalized recommendations and predictions in the context of business.

Is the tech really capable of that level of valuable input ?

First you need a strategy

“You still need to have well-defined strategies set out – meaning people, not machines, will need to come up with good questions to ask of datasets,” explains Morgan Stewart, co-founder and chief executive of strategic email marketing agency, Trendline Interactive.

Then a good question

“A college professor once told me the key to good research is asking better questions than anyone else,” he also states. “Once you have a good question the research part becomes almost formulaic.”.

And finally Data

“It can certainly help you get to the answers faster, which is great! But the heavy lifting is still needed on the data integration side. It can crunch the numbers once it has them but someone must first provide that data correctly, which is a pretty laborious task. There are very practical applications but the expectations of AI generally are still unrealistic right now.”

Roadmap

Both the tools, Einstein Prediction Builder as well as Einstein Bots, are currently in pilot and will be generally available in summer of 2018. Salesforce said pricing for each Einstein feature varies as some are already covered under the existing license while others require additional charges.

• It’s to be seen as to what extent Salesforce manages to reduce the complexity with creating bots and bring an element of underlying intelligence, but as the firm’s vice president Jim Sinai said, my
• Einstein is “automating data science under the hood.”