Have you ever cancelled plans to go to the movies because you got tired of trying to persuade the computer on the other end to get you the right reservation? Or have you yelled at your navigation system in the car because after the fifth attempt to pronounce Moenchengladbach Dagmar is still suggesting to drive you to Munich (for historical reasons I call all my navigation systems Dagmar – but that is a different story for another time)? Especially earlier versions of such solutions were limited in their ability to understand commands. You had to learn a set of key commands and do your best to pronounce them as clearly as possible. Yes – No – Enter Destination and so on … True human speech with all its facets, flavors, slang, personal style and dialect has been a challenge for engineers but technology has evolved fast.
When you look around and see the interaction with many virtual assistants these days, it feels more and more like a true conversation. With the convenience that Alexa, Siri or similar solutions offer, many consumers are willing to accept that devices are now basically eavesdropping on them around the clock. I sometimes picture a dark bar where spies from various organizations jealously listen to a software engineer who closes his presentation with: “… and we even got them to pay 180 Euros to let us spy on them …” Trust and integrity will be key aspects for such offerings – in addition to being smart, helpful and likeable. At any cost, providers must ensure that the consumer believes and trusts that the information gathered about them is not misused.
The examples mentioned so far are all consumer-facing. With SAP CoPilot, SAP now introduces a digital assistant for the enterprise. Like its consumer-facing cousins it combines native device capabilities with natural language processing, but adds additional capabilities to the package. Using Analytics and Machine Learning, SAP CoPilot can offer contextual information and recommendations instantly. Starting with a set of base rules it can recognize patterns and “learn” from behavioral data without being explicitly programmed to do so – thereby continuously increasing the quality of results. So, SAP combines a conversational user interface with an awareness of your business context and can suggest quick action, while continuously learning and improving the quality of recommendations.
In a demo scenario, you can see how with a natural dialogue you can tell SAP CoPilot that you would like to take a day off next Friday. After confirming, that you would like to create a leave request for May 12th SAP CoPilot automatically creates the document and sends it for approval (after having checked behind the scenes that you still have enough vacation days and that it is not a public holiday).
Thinking this only one or two steps further, you can imagine a dialogue with a customer where SAP CoPilot monitors the dialogue and pushes the right information to you instantly – previous orders, potential products required, availability and lead times, current promotions or information on possible payment terms. With this insight, you can not only focus on the actual conversation with the right answers and suggestions at your fingertips, but follow up activity like offers or orders can be prepared directly while you speak.
To me, voice recognition, natural language processing and digital assistants are likely to significantly change user interaction with business applications in many areas in the near future. With Sapphire coming up, I encourage you to check out and learn about the cool solutions that SAP is already delivering in this area.