We took some time to catch up with our top international Robotic Process Automation (RPA ) developer, Karol Mielnicki out of Poland.
Karol, thanks for being with us today. We’d like to start off with learning about your tools of choice.
1. Tell us about the different RPA products, which is your favorite and why?
UIPath is great because it is programmer friendly, is updated frequently and their support is excellent. It comes with a huge developer community which means more tools and libraries, thus you need fewer external tools to use it. I love Automation Anywhere as well because they have tremendous scale and a built-in tool for intelligent automation; that is unique in the marketplace.
Python and VBA scripting, Excel and GoogleAI Cloud are other key tools that I use to build robots.
2. Where do you see these vendors going? What are the trends in terms of new features, development, etc.?
I am seeing more integrations with outside tools and libraries (e.g., GSuite RPA integration enables easy updating of excel spreadsheets), analytics (e.g., ROI, manual hours saved, success rates) and intelligent automation (e.g., machine learning). RPA machine learning modules can recognize invoices and receipts. Custom machine learning modules from GoogleAI Cloud can do document categorization (e.g., “this is a car loan,” “this is a tax payment”).
3. How many bots does the average customer have deployed?
My smaller customers have 3-4 bots deployed and larger customers who have been in the game a couple of years typically have 100 or more bots.
4. When they have 3-4 bots, what do they need to support it?
They need to work with IT to ensure that they know RPA is in production and that they have people who understand the robots and can receive support tickets. The support resources know who the developer was and can reach out for more complex issues. Note that there are ways to write the code so that fixing the bots is easier; this is a key dependency for the supportability of bots.
5. When they have 75-100 bots, what is needed there?
With this scale, a company needs a Center of Excellence staffed with process experts who manage a portfolio of bots, platform administrators and RPA developers. Note that you can outsource the technical components, but you should contract with a company that provides RPA-as-a-Service so that they will retain people who know your bots and your environment. A company using RPA at scale also needs to constantly market the program as there are unique organizational factors that can work against an RPA program (e.g., you don’t want to have a reputation for doing something that is viewed as reducing the need for humans and taking away jobs).
Be sure to read our blog about a company that scaled to 200 bots:
"Customer Conversation: Scaling Robotic Process Automation in a Global Consulting Firm."
6. What is the biggest mistake you see companies making when they start out with RPA?
Lack of planning and narrow vision. A company shouldn’t go into this expecting to only deploy 1 robot because there will generally be diminishing returns unless that 1 robot is handling a very important, high-ROI, high volume task.
Selecting the right processes for RPA is critical, and when someone who is not a process expert selects the process to be automated I see them make mistakes which put the future of RPA for that company at risk. The person who selects processes should be experienced with mapping out processes, committed to artfully managing a pipeline of projects and able to plan how the robots will work together in the future. They shouldn’t go in planning to only use 1 robot.
Author's note: be sure to see our blog which also covers this topic, Customer Conversation: Scaling Robotic Process Automation (RPA) in a Global Consulting Firm.
7. What mistakes do more experienced companies make?
If an experienced company has 75-100 robots, they are doing something right! I have seen them make some mistakes that prevent scale, though. For example,
They don’t track and communicate ROI, which can put an RPA program at risk.
They don’t set up a center of excellence which causes the initiative to be fragmented and inefficient;
They fail to build a strong partnership with IT and struggle with security or program support; and
They don’t manage a pipeline wherein they prioritize requests and make sure they plan on how robots will synergistically work together.
Regarding planning, here is an example: If a robot goes in and only scrapes 30% of the data because that is all he needs, and a 2nd robot is built later that goes and scrapes the rest of it, you’re going to be missing something about the process and you’re double-developing when you could just capitalize on the first robot and add to it. In this case you should build in the sharing of information between robots for efficiency, effectiveness and keeping the process holistic.
8. What mistakes do inexperienced Developers typically make?
This is a huge list! I’ll just highlight the top ones.
Not making things reusable (see the 4th bullet above);
Failing to make the robot scalable (e.g., you build a robot for a limited use case where it only scrapes one website which requires lower volume of data, which then causes it not to work on 2 websites that are similar);
Not planning (refer to former points about reusability and robots working together); or
Hard coding data into the robot: a developer should not hard code a link or other specific variable into a robot; all code should be dynamic, for example, the robot should operate from config files or get instructions, variables or data from the orchestrator as alternatives to hard coding.
One other point: “Low code/no code,” or the notion that you want to achieve something with little coding so that it is easy to maintain, is a misnomer here. Business people typically can’t build robust, scalable robots unless they are going to become programmers. Doing this requires coding expertise and there’s no way around that given the tools that exist today for RPA.
9. What changes are you seeing in the RPA world since COVID hit?
When COVID first hit, everyone got scared and stopped what they were doing. But now there is a huge boom in RPA for 3 key reasons: reduced budgets mean that companies have to do more with less, companies are realizing robots are reliable and COVID drove a rapid acceptance of working from home.
Regarding work from home, prior to COVID, the majority of companies still were not remote-friendly for workers because they lacked security, VPN access, etc. This meant that much of a person’s work had to be done from the office (which reduces the number of use cases that can be handled by a robot). When COVID hit, these companies had a huge problem and they filled these gaps quickly. Now that most work requiring systems can be done virtually, there is a much larger number of processes that can be automated. In other words, better security and VPN access are dependencies for many process automation projects.