posted

0 Comments

Throughout the past year, VMware has embarked on a mission to reimagine the support experience. As part of that mission, we have built a software engineering organization within Global Services, focused on developing great software to create a proactive, personalized and effortless experience. One of our first steps in that journey was the development of VMware Skyline, which we launched at VMworld last year. Now it’s my pleasure to give you a view into one of the next areas we are leaning into.

We’ve spoken to many customers over the last year to understand which areas we are best at and where we could improve. It’s clear from these conversations that our customers generally love our technical support engineers and get enormous value from their expertise. What is also clear is that we haven’t done a good enough job getting our customers to the right expert as quickly as possible. Sometimes we ask the same core questions, sometimes the support request seems to bounce between multiple people until it finally lands with the right expert for the issue who can usually solve the problem quite quickly. Using the powerful capabilities of Watson Natural Language Classifier, Watson Assistant, and Watson Machine Learning, we will be shortening that cycle and getting the issue to the best person to solve the problem as quickly as possible.

First, we know there are common pieces of information that are almost always needed to understand the issue and solve it as quickly as possible. But those pieces of information could vary somewhat, depending on the situation. We also know that pages and pages of drop down selections is not a great experience. To make this simpler, we will be using Watson to interpret the natural language description of the issue and discover those pieces of information within the text. Visual cues will light up when Watson has found a phrase that matches an important category so our customers will have a clear indicator on whether the description is robust enough and can also reject or replace those matches. Including all of the information isn’t always mandatory, but we do have data that shows that the more categories that match, the faster we will be able to solve the case.

After the case is submitted, Watson uses the information in the case and evaluates it against the history of previous cases submitted across VMware to predict the best match. Our software can look at criteria like personal case history, issue type, products, versions, error strings along with our internal metrics like closure rates and training to find the best expert for this specific issue. The information we use within the model will continue to change as we get more data and learn new patterns, but we’re already seeing great improvements in our ability to recognize and predict the right expert.

Once the case is handed to an expert, Watson will also help that expert get the issue solved as quickly as possible. Because we are able to evaluate similarities in our classification models, we can also guide our support engineers toward any previous cases that may be most closely related to this one. That should provide a great head start in their research. We will also be able to find opportunities for new knowledge articles based on these common issue and resolution patterns.

I’m very excited about our work harnessing the power of Watson to improve our support request experience. While we’ve done a lot to prove our this approach, we still have a little ways to go until launch so expect to see these changes beginning to emerge in early 2019. We also have a few other transformational efforts in the works to incorprate the power of Watson into our Global Services business that we will be able to share soon. Thank you to all our customers for your continued partnership on this journey as we continue to reinvent VMware support.