A smart vending machine

The second week of November I will be at TeliaSonera M2M symposium and the Swedish TechDays. I will be bringing some of Avanade’s thinking in the retail space. I will be bringing what we call the smart vending machine.

imageI won’t be bringing an actual vending machine unfortunately. But a concept that we built together with our customers.

Our showcase solution utilizes a Microsoft Surface as the touch point, an Audrino as microcontroller and Microsoft Intelligent System Service as a backend.

It sends all events to the ISS and store them in Azure. In it’s essence this is an IoT solution with all the bells and whistles. The tech is really cool from a geek stand point and I am looking forward to prepare it for the events.

But where is the real value here? The tech is cool. But what would make a business pay for it? I will focus on two technologies Microsoft have been talking about:

Both these solutions is about turning events and data into insights. It is these insights that gives our vending machine the potential of being smart.

Getting insight in event streams

A machine sharing it’s temperature over and over and over again are usually not useful. I want to know when the temperature is below a certain threshold. With stream insight I can query the stream of temperature readings for an abnormality and can react when there is an issue instead of monitoring the actual temperature.

There are other things a vending machine could query for; abnormal transactions that could be fraud, unusual amount of transactions that could disrupt the logistics. And so on.

Predicting the future

By storing all the events about sales, environment, clients, transactions; Machine learning can draw insights about the data and try to predict the future. This is extremely useful when predicting refill. The image below shows how the vending machines inventory levels would look like in a months time based on past sales.  image 

The new smart is based on insights like this. This is where the true value of internet of things and BigData come into play; When you can convert it to value. The new services in Azure will make this happen.

When the services this solution is based upon are made publicly available I will try to post a few examples on how to realize what I’ve discussed here.

One thought on “A smart vending machine

  1. Pingback: @WinObs Tweeted Links for November 30, 2014 | WindowsObserver.com Wiki

Leave a Reply

Your email address will not be published. Required fields are marked *