The robots are coming! In fact, they have been here for a long time. But this time more advanced and intelligent than ever. In the near feature, robots are expected to overtake jobs from unskilled workers as well as people holding an academic degree. As a business intelligence consultant with a great interest in software robots, I was wondering how far away we are from having robots interacting with reports and dashboards in stead of human users. I decided to create a dashboard and build a software robot for using it. My plan was then to stress out the robot and see how many disturbances it can handle before it gets confused.
I designed a typical interactive sales dashboard for financial reporting, where graphs and visualizations are used as filter functions to drill down into the data.
Then it was time to decide which tasks I wanted my robot to perform. I programmed it to select the consumer segment using the buttons in the upper right corner. I also wanted it to drill down and look at data within the furniture category. This is done by clicking the blue part of the donut chart. Finally, I decided that my robot should read the number in the middle of the speedometer and insert it into a spreadsheet. I can’t deny, that I felt a little like Victor Frankenstein, when I saw Debbie, the Dashboard Robot opening my browser and navigating my newly created dashboard for the first time.
At this point, I guess you may be thinking, why on earth I insist on calling her Debbie. Meanwhile, research has shown that people tend to get a closer relationship to robots, when they are given a name and thereby a small step closer to an actual personality. That makes people more forgiving when the robot fails. On the other hand, it also makes people show sympathy and compassion. Sometimes so much they want to protect the poor thing from any harm or damage that may occur, so bear with me while I explain how I tried to confuse Debbie to test how robust she really is.
Obviously, dashboards change over time as the data behind them change. To replicate that, I started changing the number of transactions in the furniture category. This to increase the part of the donut chart, Debbie uses to filter the dashboard. I also changed the data values so the average price got 3 digits in stead of just 2 even when the data was filtered out. She barely noticed the changes, but used the dashboard, as if nothing was changed. To give her another challenge, I reduced the size of the buttons for selecting customer segment as well as the data color on the donut chart. Even that did not affect the navigation and filtering. As the evil robot developer I apparently am, I, of course, needed to force Debbie to give up. It was not as easy as I expected. In fact, I got to change the color of the average price to a cramped yellow, that was difficult to distinguish from the background before she returned an error.
All in all, Debbie has the potential of becoming a quite helpful assistant in the daily work of different companies by her ability to interact with interfaces designed for human users. Indeed, it may seem stupid to make a computer imitate a human workflow, instead of just loading the data directly from my dataset. However, that is unfortunately not always possible. Imagine you need to collect all your competitors’ prices on specific products. It is hard to imagine them sending their updated price list to you every month. Then Debbie comes to the rescue but is that a stupid way of reaching the information or will we in the near future be creating IT-systems with an interface for users and one for robots? Feel free to leave a comment on this.
Play the video to see Debbie in action.
You can try out my sales dashboard using this link:
Fun facts: The dashboard was created using Microsoft Power BI and the software robot was created using UI Path.