A good example of using AI technology to perform a specific, well defined goal as the start point for further development can be seen in the way some businesses are using Moveworks, a cloud-based AI platform. At its heart it is designed to both manage and provide a solution to all those level one service tickets that beset the help desk of every business – the classic ‘how do I change my password’ issues that confront every user now and then and, not to put too fine a point on it, drive support staff nuts.
At first sight, these look like tasks ideally suited to automation, but in fact rarely are, if only because users often end up adapting their own operational needs to match the way the automated process says you can do things. When the third-party process is more important than what is required by the business, there will always be problems. This does, however, make it a good target for the application of some well-aimed AI.
If that AI then has the potential to adapt to what type of service tickets a business needs, and go further by being flexible and adaptable in the range of support services it can provide, it can not only increase the productivity of the support team, but also make what can be one of IT’s more tedious jobs a good deal less so.
This has certainly been the experience of Russ Harris, Vice President of Operations with AppDynamics, an Applications Performance Monitoring vendor recently acquired by Cisco. This has some 2,500 employees and despite the company’s technology orientation, Harris is the first to acknowledge that not all of them are IT experts. But with an annual turnover growth rate of around 30%, they are all working under pressure, and therefore have high expectations and, like most people, can find it very frustrating to find themselves shut down, or have their productivity choked off for some technical reason:
So we made the conscious choice of investing in providing ‘white glove’ support for our employees, because it just makes their work life that much more productive and that much more satisfying.
This is particularly the case for the staff providing the support. The company has a staff turnover metric that is a specific strategic objective of the company. The goal is to drive the retention rate up and the provision of a ‘white glove’ support desk environment has been a key part in making that work.
For ‘white gloves’, read Moveworks
This environment, which has now been in place for around a year, is built on the Moveworks platform, which combines AI with natural language processing to understand and interpret a user’s problem, and AI to understand that problem and identify appropriate solutions. It is built on top of a foundation of learning that came from working through some 20 million tickets from a handful of large business users. So it has developed a pretty good selection of stock answers to many of those low level issues that users have.
But the AI element brings with it the potential to expand the number of services it is able to deal with so it can, for example, not only interpret a badly phrased support request, but also learn how to cope with ever more complex problems, freeing up the support team to address more productive and fulfilling work.
One of its other advantages is that it also runs 24/7 allowing it to match the work patterns of most companies, as Harris observes:
If I happen to be doing some work at home on a weekend and I need to get an updated copy of an application, I can get a updated copy loaded, on a Saturday night if I want to.
The use of natural language also provides a level of flexibility in how the system can be used, for Harris has developed a ticketing approach all of his own, that developed out of the way he tends to work:
As the VP of Operations I wander the halls and people frequently come to me and say, ‘Did you know that this was broken or that conference room has got an issue?’. And, you know, I don't have all those things at my fingertips. So I have become a frequent submitter of tickets into the system, and it's easy for me to just ask the application’s ‘AppDynoBot', as we refer to it internally, to show me all my open tickets.
There is, he says, an easy interface to work with which can provide useful summaries of the tickets, so he can query which ones are open. Then he can follow them up, drill down into them if required, and communicate directly with the support team, all without having to access the ticketing system on the company’s Atlassian Jira software support team system. It also interfaces with the Slack instant messaging and staff collaboration tool, providing Harris with the ability to not only message appropriate staff but also attach relevant documents and files.
Increasing what the ‘white gloves’ can do
Moveworks does demonstrate a clear AI benefit, even if it does require some manual intervention – indeed, that intervention can become one of the more creative and fulfilling roles that support staff acquire along the way. That benefit is the ability to grow with the demand for extra capability. The company’s approach is to stay involved with its customers as far as possible, and AppDynamics is exploiting this to increase the productivity of the ‘white glove’ support service.
The AppDynamic team meets regularly, at least once every other week, with a team from Moveworks to analyse the full ticket history and identify all those tickets that were not able to be resolved by the standard merge process. This merges ticket data with a knowledge base of remediation and solutions data, most often in the form of articles written by the combined team.
As well as the knowledge base AppDynamics has also given the Moveworks system the org chart of the business. This gives support staff information about the relationship between individual staff members, and is particularly important when dealing with new hires and their reporting lines.
Data from an unresolved ticket is then analysed against the suggested solutions and, where necessary, the available articles can be amended or updated. The idea behind this is simple: get the information delivered to the user so that acting on it solves the problem and it becomes an addition to the list than can be solved without support staff intervention.
AppDynamics went live with Moveworks a year ago, having spent some 90 days from signing up for it with the initial provisioning of the knowledge base and getting the Moveworks environment integrated in with its existing systems. Harris recalls:
I remember early on we were able to get between 10% and 15% of the tickets automatically handled. And then, over time as we have these meetings and we add more richness to the knowledge base, we have got to the point where we now have upwards of 30% of them being handled automatically.
We've been having conversations with Moveworks to help them help us understand what they believe is achievable and how to drive towards that. Internally, we would all be very happy if we got anywhere close to the 50% mark of tickets being handled automatically. But I know that Moveworks has aspirations to go as high as 70%. I'm 100% supportive of helping Moveworks get to that point, and we will continue to run this regular process of enhancing the knowledge base. But internally, we built our business case on 30%. If we get to 50%, that would be a home run in my mind.