Analyst Corner

Artificial Intelligence: The 30 Year Success Story

I can confidently say that my very first exposure to artificial intelligence happened about thirty years ago. I was strolling through the computer engineering building on campus at Georgia Tech and happened onto a computer screen facing out onto the hallway. On the screen was a changing array of symbols seemingly displayed at random, catching my attention. Being the curious college student I stopped to look and as someone walked by, noting that the screen was clearly on display for all to observe, I asked what the high tech mumbo jumbo was all about. It must have been a graduate student because they replied rather nonchalantly that it was an artificial intelligence experiment and continued on their way. Sensing in the response that it should be intuitively obvious what AI was and was all about, I continued on as well, not giving it much of a second thought beyond the times I would make my way through the building.

Fast forwarding about 30 years and AI is all the buzz. These days be it in a technical journal, user conference or even watching television and catching an IBM commercial on how Watson can help me with my taxes, it’s hard to escape the growing presence of AI and the consideration of how it is and will impact our lives. At this year’s Salesforce XChange event (formerly Demandware XChange) AI was front and center as in my opinion, the macro theme of the event. Salesforce brands their distributed AI as Einstein, and marketed through a very non-threatening and disarming cartoon character depiction of Albert Einstein.

AI for all of its variants comes in three general categories from IHL’s perspective. First there is the approach of IBM and Watson. While marketed through their participation in Jeopardy and other commercial products such as tax software, when you peel back the layers as we are told is really a series of APIs wrapped around custom services and designed for your specific application. The second variety of AI, would be solution specific and purpose built applications. There is a whole litany of these and more introduced daily. We like to think of them as the next step in the evolutionary maturation of Business Intelligence and Analytics tools, think BI on steroids. The third category of AI solutions and the place where Einstein fits is as functionality built into an existing application that is mostly transparent to the user. Think AI, with the analytical benefits, but with the simplicity and inviting demeanor afforded a cartoon character.

At XChange Salesforce highlighted general availability of Einstein across seven products. These products encompass both what are core Salesforce offerings along with the e-commerce assets acquired through the Demandware acquisition, and the marketing assets acquired from ExactTarget.

While there was really no discussion around the previously announced IBM Watson partnership and what that means, the introduction of seven specific AI application offerings was proof that AI was a clear part of Salesforce’s product strategy moving forward. And while I make somewhat tongue in cheek comments about cartoon Einstein, the key message conveyed at the event was not that AI was some mystical tool only to be accessed by seasoned data scientists, but a tool available and easily usable by any Johnny or Janet user.

I make the point here to highlight that this is something they absolutely got correct I believe and their approach to integrating AI into the core product has the potential benefit of quick returns and high adoption. I also believe there is an aspect of this that is a clear differentiator in the marketplace because it means that anyone can utilize the benefits of AI and come up to speed very quickly.

The other point worth making on the subject is that this wasn’t vaporware or a single development user, but they shared real results, from what I understand to be eight clients that are already using their predictive sort capability. By way of background as the shopper embarks on the shopping journey, predictive sort will make product suggestions based upon your search and purchase history. In talking with fellow analyst Nikki Baird of RSR Research she noted that their research showed typical per customer revenue lift in the 4-8% range. Data shared at the event put that number for the eight trial customers at an average value of 9.1%! Not too shabby at all especially for a beta product.

While it is certainly an overused phrase I really believe the sky is the limit with AI. AI in retail software as a whole is very early, adoptions sparse, with limited use cases for application specific packaged software in the retail marketplace. Furthermore we’re at a place where only a handful of early adopters have AI baked into core solutions. The point to think about though is to consider that the early returns are far better than historical returns. What this means is where the ROI bar is high and where retailers are looking for any way possible to stretch their IT budgets, results such as were touted at XChange might really help move the AI adoption needle more rapidly than the typical retail technology adoption curve. And while it only took me 30 years to see the fruits of what AI is, early returns suggest it was definitely worth the wait.