#PaiRED: It’s Still Up to Humans
Ultimately, it still falls to humans to define the workflow up front. Humans are also in charge of process recording and writing and editing scripts. Sometimes IT sets up the flow and sometimes flow-chart-like choices are available so that “digital citizens” can develop the process flow. The most difficult aspect of reproducing business processes for automation is defining what the business processes are to begin with and why they exist. Sometimes process mining tools can help with this and sometimes it has to be done manually.
Choosing an RPA product and vendor is not easy. All vendors have priority file structures and there are no standards for RPA. Careful evaluation up front and doing a proof of concept can save pain and money. Before committing to anything do the homework because changing your mind can be a disaster and be very expensive. Look at the features and look at the work environment without rose-colored glasses. Build sample scripts and create a proof of concept for as many possible scenarios as is realistic. Don’t just tackle the easy stuff. Use the tools that are provided and make sure the orchestration works. Go through the entire process flow in detail and then do it again and again. There needs to be a wide range of different personas tested and a low code environment for developing bots and business rules. Finally, it might be helpful to have professional programmers look at how they might write real automation code to call the RPA tools APIs. Then ask the hard questions like how it handled exceptions and how much human review was really involved.
Generally, these bots will work off of limits and probability rules. Somewhere between those boundaries is room for human judgment, and it is important that the RPA tool recognizes this and is capable of submitting the case for human review. It’s very important to do the work up front. If an organization selects the wrong vendor, it can be disastrous. If the RPA system isn’t implemented right and needs capabilities it doesn’t have or are missing, it’s not easy to switch. Testing and process evaluation up front and in depth are essential. Documenting each step of the RPA process including creating each bot might help. Remember, with RPA, not AI, humans are in charge of defining the flow.
New vendors for RPA are springing up all the time. The challenges for organizations are usually internal. RPA will facilitate the automation of flow between existing infrastructure and systems, but humans have to figure out what that is. The main reason for mplementing RPA is to maximize profit and minimize workload. Implementing RPA can become a large-scale, long-term nightmare if the processes and testing are not done well.
Another challenge to RPA’s success is the systems it tries to link together automatically. It is critical the business has systems that can uphold the RPA tools so that accurate output can be generated from them. Timely updates and monitoring are necessary because the technologies move quickly, and RPA should be able to withstand changes. RPA failures can come in many forms like bot failure, system error or unexpected random issues. It’s important to have someone monitoring the RPA and keeping an eye out for systemic failures.
Finally, another issue is using RPA where RPA really isn’t the answer. It is very important to review and audit RPA. Sometimes RPA is assigned to a non-repetitive task leading to unnecessary costs. The most formative challenge to successfully implementing RPA is identifying the kinds of tasks it will do. Next is identifying the tasks and the process flow for each option. There are research-based indexes that calculate a score for RPA suitability. They provide guidelines to organizations making the transition to RPA. Processes with high workload and low complexity are good candidates for RPA. The business function and the industry are also important. If a business process is easy to define, standard and repetitive, and follows well-defined rules, RPA might be a good fit. If it is time consuming and requires manual intervention with a computer screen interface, it is probably a good candidate for RPA.
#AI@Work, #WFH, #WorkFromHome, #BobbeGB, #BobbeBaggio, #ThePajamaEffect, #Touchpoints, , #Visual Connection, #PaiRED
