During the last week of January, a colleague and I visited the LEAP (Lead Enterprise Architecture Program) conference at Microsoft’s headquarters in Redmond (WA). For 5 days we were updated about the latest technologies and we got a look into the future as Microsoft sees it. From a technology point of view, it is great to see what you can accomplish with all those Azure services. But, in the end, you probably want something in the core process of your business to go smoother, more efficient or more cost effective.
At LEAP, Microsoft talked a lot about containers and microservices with Kubernetes, Machine Learning and cognitive services. All great technical offerings, on top of which you can build nice applications. But why should you? Your company is building great applications and has been doing so for a long time. You might already have adopted a cloud first strategy and moved everything you wanted to the cloud using some form of SAAS, PAAS or IAAS. In the meantime, machine learning and cognitive services have come along, and you are wondering if you should use it and how?
This might be the technical side of your story. On the other hand, your business is more demanding every day. Business users experience many new features in their private life. Their phone tells them how much time they spend on certain activities. They are playing games with their HoloLens and are talking to their phones and speakers to get updates on traffic, weather or the latest fashion trends. For them, the future is now and they want to experience that in their business lives as well.
Some people will read this and think; ‘not my business users, they have never asked anything that fancy’. You are right. They have not yet asked for it. Most business users are busy getting their daily work done. Yet, you need to be ready once they will ask. But, be ready for what?
The combination of Cognitive interaction and machine learning (sometimes overenthusiastically called AI, which is something far more complex and intelligent) will make future solutions more people centric in a way that they suit users’ needs by behaving more ‘human’. People expect to talk to their devices just as they talk to you. They expect applications to give them good advice based on events in the past as any other colleague could give them. A change that forces your IT organization to think about data quality and differences in the interpretation of natural language and gestures instead of keyboard and mouse interactions.
Why should your business users fill in metadata, to make the files they upload easily retrievable? A machine learning application can do that faster and, after enough training, more accurate. (Functionality already offered in, for example, an E5 Office365 license.) Why should they call a busy service desk, wait for 5 minutes and get an agent on the phone who searches for the answer in a script or database and hoping they will find the right answer the first-time round. Two challenges easy to visualize.
Now take those examples one step further. Visualize a small medical production department producing fabrics on such a volume not worth automating. Nowadays people must sign in to a log if they enter such a facility and sign out if they leave. The person who plans the activities tries to be as efficient as possible while keeping workers happy by offering enough variety in their tasks. The log for that production department could easily be replaced by a camera (that probably is already there for other security purposes) which is extended with facial recognition and a small database logging every time an employee enters and exits the facility. It would minimize the risk of someone forgetting to sign the log. Machine learning could suggest the best planning maximizing efficiency and employee happiness. It would also lead to serious privacy debates as well as further regulation that is for now, beyond the discussion of this topic.
‘But for those platforms that offer machine learning, we still need to train their models with lots of data? Isn’t it?’ Yes, you must. However, Microsoft now let you use, and if you want, even extend, their models. If you leverage those models with cognitive services already offered, you will have a great baseline for your first real people centric application.