We were a couple of months along our robotic journey and although we had our family of robots, who are able to do one action within a process, we still hadn’t quite managed to create that elusive robot that was able to complete a whole process, just like one of our team members would.
We were still in a bit of a discovery phase, where we were trying to work out the best place for our robot. Now although this may seem like quite a long time to be looking at processes, it certainly didn’t feel like time wasted. One of the additional benefits of doing this work was that not only were we discovering things for our robots to do, we were also discovering things for our teams to do – lucky them!
With a LOT of potential processes and not as much time and resource as we would need to tackle them all, we needed to be ruthless in what we agreed and didn’t agree was fit for RPA at this stage. If there were processes that needed a human touch because they were complex and full of ‘ifs, buts and maybes’ when it came to decision making, or it didn’t bring any time or cost savings for a robot to do the work, or the robot could only do part of the process, then it was humans 1, robots 0. Our learning was that you should only spend a lot of time and money developing a complex robot to do everything if it made sense for them to do so. See, we told you people were still important, robots can’t do it all – at least not at this stage…
After having stared at an overload of processes and painstakingly working our way through each of them, we could see the light at the end of the tunnel and that light was shining like a spotlight on Teachers Pensions. Could this be our holy grail of processes that allowed our first robot to come to fruition who could complete an entire process on its own? We were excited. This is what we had been working towards and now was our chance to make it a reality. Building and developing started moving full steam ahead, nothing was going to stop us now… until the system freeze kicked in!!! Were we NEVER going to get this robot live?!
The system freeze gave us no option but to down tools. Time was precious to us but if we were being told that essential maintenance was happening to one of the systems that we needed to be able to build our robots then there literally was nothing we could do about it and the Lab came to a halt. Bearing in mind that we had removed team members from the business specifically to work on RPA, this wasn’t ideal. I’d also love to tell you that we had planned for this and there was a contingency in place should such an issue arise but alas, no, we hadn’t been aware that a freeze was planned. FAIL! Another thing to add to our learning!
We spent the next couple of days utilising our time by planning our next steps (and building in some contingency)! Oh and praying for the freeze to thaw out! Our deadlines had now slipped and we needed to get back on track as soon as the freeze lifted which, luckily for us, happened a couple of days later. Go, go, GO!!!!!
We finally got to a stage where we believed that our robot, who we lovingly named Nibbler because he nibbles away at the backlog was ready to be tested. The idea was that he followed what we called ‘The Happy Path’. The Happy Path is what a person would do, brushing aside all of the system pop ups, ignoring the different layouts and skipping missing records. All of the decisions that we, as humans, take for granted. It was the moment of truth…
How do I politely say that the communication between robot and system was not quite there? You know when you go to a foreign country and they speak a different language to you, but you try to explain what you want from the other one with lots of gesticulating and in very broken lingo, where maybe one word in every ten is the right one? Well this was a similar affair! Nibbler had been built to follow a certain path. However systems don’t always like to offer the same path and might occasionally throw in something new or unexpected. In fact our Developer Phil James summed this interaction up beautifully in the following dialogue:
Robot: I’ll just click on this button to get to the next screen.
System: Have a pop up box.
Robot: I’d better click the button again because it’s still not loaded, maybe I didn’t click it hard enough.
System: Erm…pop up box. You’d better close it first.
Robot: Hello! I’ve clicked the button again.
System: I’m still waiting for you to close my pop up box. I’m going nowhere until then.
Robot: Click, click, click. Why is nothing happening?
You might also like
Robotics Software: EXCEPTION THROWN.
So we put in a check for a message box, in case this reoccurred, and tried again. After five minutes of successful automation…
Robot: I’ll click on this button to get to the next screen.
System: Here’s the next page.
Robot: Enter this data here, then navigate to the next page.
System: I’ve just decided that you need to put a tick in a box on this page. I haven’t asked you to do this before. I love being spontaneous.
Robot: Enter some more data. Hang on, where’s the box I need to fill in? Help me!!
Robotics Software: EXCEPTION THROWN.
So it appears that building robots really is becoming quite difficult. We’ve experienced unexpected system freezes and extreme communication breakdowns, are we ever going to master this? It’s back to the drawing board we go…
Want to find out if we finally manage to get a robot up and running? We’ll reveal all in our next monthly blog.
Missed previous blogs about our journey into RPA? Find them all here: