Enter the Next Level of Customer Support Experience with AI Technology

Jens Trotzky explains AI-driven support
SAP’s Jens Trotzky explains AI-driven support.

AI-driven product support takes the SAP customer support experience to the next level. Jens Trotzky, head of Artificial Intelligence Technology for SAP Support, explains what AI-driven support is, how it works, and its benefits to SAP customers.

Q: What is AI-driven support?

A: AI-driven support is SAP’s next step toward a more automated, personalized, and preventative customer support experience. Based on machine learning technology, this new support approach will help to provide relevant answers much faster while continuously improving the service with each query being processed and finally help prevent incidents.

How does it work?

Over many years, support has been a very data-driven exercise. Finding the cause of a problem — the so-called root cause analysis — heavily depends on data to narrow down an issue and identify a solution for even highly complex problems. Machine learning technology is centered around pattern recognition in Big Data sources. In AI-driven support, we put both elements together to analyze large amounts of support data and provide real-time answers to customer problems. The use of machine learning allows us to be case-specific and avoid a one-size-fits-all support approach. The user is now in the center of a fully personalized support approach.

Why will this change the customer support experience?

In the past customers had to open a ticket, which in return was read and analyzed by a support engineer to finally recommend a solution from SAP’s pool of knowledge and documents. This process can be time-consuming given the high volume of support tickets and existing knowledge documents. With machine learning capabilities, solutions can be significantly narrowed down. By using machine learning technology it also takes only a few seconds for the first recommendation. Therefore, in the future customers can create a ticket and get a first suggestion for a solution within seconds. However, should the solution offered not fully answer the question, this feedback will be sent along with the ticket. This allows for looking for better solutions much more quickly and to improve machine learning algorithms to provide even better solutions with every search added to our data pool.

Does this mean human-to-human interaction will no longer be necessary?

Not at all! The way and the content of human-to-human interaction will change and become much more sophisticated. Machine learning can help classify and solve many questions and will complement the support process both on customer side — as well as on the SAP engineer side, with the overall goal to speed up the time to solution and further improve the customer support experience.

Where are we today? How long will it take to fully unfold the benefits of this new technology to customers?

SAP recently launched the first stepping stone in this journey, making available incident solution matching to all customers using SAP ONE Support Launchpad, SAP customers’ first entry point for all solution-related support questions. This allows customers to create a ticket and receive recommendations while entering technical questions into the query. Even while the ticket query receives recommendations, customers can continue to send the incident to SAP product support. Next features are planned be released throughout 2019, but we envision a time frame of two to three years to provide even more personalized predictions that manage to go all the way to the customer system level.

What AI-driven support solutions and services does SAP offer today?

Today, incident solution matching is available to all customers of SAP One Support Launchpad. For the support organization, in this context, one of the biggest challenges for using natural language processing is an industry’s terminology and specific language. Typ­ically, many words are used across entire product portfolios with sometimes completely different meanings — such as “customer,” “material,” “order,” “container,” etc. — which requires accurate understanding of the overall context.
The first step in realizing AI use cases on a comprehensive roadmap is SAP’s incident solution matching, a service that matches incoming incidents with previous solutions, thereby enabling users to get technical support for known problems much faster. Making use of millions of anonymized past incidents, the feature matches the most relevant answers to any given technical question. This is certainly a Big Data application. Since its go live in November 2018, customers can benefit for the first time from AI-infused interactions inside SAP’s core support pro­cess, brought to users through SAP ONE Support Launchpad. SAP tested it with several hundred engineers over months and found the results to be very accurate for products where a lot of his­torical data is available.

What comes next?

AI is very different from previous technologies. Machine learning-based services in support are heavily dependent on mastering available data. Therefore, SAP Sup­port founded its own team of AI experts to make quick progress in highly sup­port-specific innovations that increase the customer support experience. Our vision is to provide proactive support to customers using predictive analytics. To build credibility, customers must receive only very few messages that are highly relevant to their individual business setup and processes.

After the successful external launch of our incident solution matching service, we are already preparing the next use cases. Internally SAP is currently testing incident-to-incident matching, which will be another highlight we hope to present to customers soon. We are also working on support entity recognition capabilities that will allow SAP’s support AI engine to be even more accurate in the future.

So stay tuned, there is much more to come!


Sophia Stolze is the communications lead for Customer Experience Product Support at SAP.