We now know the big product announcement at next month’s Dreamforce will be “Einstein”. This has been the culmination of internal development projects and a number of acquisitions over the past, well several, years. We know very little about Einstein other than a few teasers alluding to artificial intelligence we can make a few guesses. Let’s face it, there’s no fun if we just sit back and wait.
Reported in ZDNet, Marc Benioff made the following comment during the Q2 earnings call;
Companies will demand that your software is going to be intelligent, smart, that you’re going to have machine learning and deep learning and machine intelligence built-in. Machine intelligence is going to be declared as well as programmatic…I really believe we’re going to have the best artificial intelligence platform in the industry.
Based on this quote there are 3 underlying related technologies to Einstein, machine learning, machine intelligence and deep learning. What does this mean for the average Salesforce customer ?
- Machine learning is the ability for a computer to learn without specific programming. We don’t program the computer how to solve a problem, we program it to learn how to solve problems. The foundation is pattern recognition and computational learning, understanding trends or patterns and learning from them over time. For the Salesforce customer this could mean predictions based on trends in your data. Instead of staring for hours at dashboards and reports trying to draw inferences from the data Einstein could provide those insightful gems of prediction for us about why our ice cream sales tank in the winter (OK, anyone can guess that one).
- Machine intelligence provides the tools for a computer to perceive it’s environment and make rational decisions to increase it’s chances of success for a given task. The concept Machine Intelligence has been used in Terminator or Matrix doomsday scenarios but it’s more likely to be channelled into niche uses. We interact with this emerging technology daily through Siri, video games, purchase protection systems and websites that attempt to predict your needs to secure a purchase. Could Machine intelligence helping to decide the best sales strategy based on historical performance is an interesting idea.
- Deep learning has previously been referred to as a neural network and provides a multi-layered approach to understanding a problem. It’s been used extensively in photo recognition which needs millions of examples to differentiate weather, night/day, cities, people, etc. With each photo the system learns a little more adding a more accurate decision making capability. For a Salesforce user this could mean the ability to scour social networks, sifting through good and bad feedback, dealing with issues as they arrive. While we already have this technology I’d argue it lacks the finesses to differentiate the nuances of human language and intention.
Does this mean we can sit back and let the machines take over? Probably not yet and definitely not for some time. But the ability to use AI to assist decision making is certainly a welcome addition to a great product.
From what I understand this probably isn’t a separate cloud but an underlying technology foundation that supports all clouds, probably starting with sales.
I’ll make sure that I update this post when the announcement is made.