Don't go for crappy Robotic Desktop Automation and select the right vendor

The purpose of this article is simple: provide a high-level overview of RDA software spectrum. The objective is to ensure that you can make an informed decision when selecting your RDA technology (or at least ask the right questions).

Hopefully the picture below provides a good summary.

I will skip the “trigger” part as this is quite self-explanatory and will only focus on the different types and attributes of RDA.

Rudimentary type (a joke)

This RDA type is simply the most primitive and probably one of the worst automation method. You should definitely stay away of this unless you have a very valid reason. With a synchronous engine, the employee sends a request to a bot and must wait for the completion of a task before she/he can continue with other activities. Given that the bot is running on the employee’s desktop, the employee cannot perform other activities in parallel. One thing for sure employees can do is go for a walk or just eat some popcorn while watching the bot. Is this actually an ethical behavior? The RDA engine is based on unsophisticated screen/ web scraping methods. Stay please stay away from it.

Basic type (a better version but still a joke)

This is where the task execution becomes asynchronous...NOT the engine but only the task processing (and this is a very important note). The employee trigger a bot that automatically run in a virtual desktop so that she/he can still continue performing other activities in her/his physical desktop. The RDA engine is based on basic GUI automation methods. This type of RDA is not very resilient and appropriate monitoring mechanisms must be implemented. Also there is an extra cost associate with this type of RDA: setup and maintain virtual machines.

Advanced type (the proper way of doing things)

Moving to the “Advanced” type, this is where RDA moves to an acceptable level. This should actually be the minimum requirement in my (humble) view. With an asynchronous engine, the employee send a request to a bot and continues processing other activities in her/his desktop before the bot completes the task. This is absolutely crucial in RDA.

This is not only about interacting with the back-end applications via advanced routines and functions. A more robust “Kernel” based engine interacting directly at the operating system level is also used in order to integrate even more closely with target applications. Kernel based engine allows to access services from the core component of the operating system. To put it more simply, the bot interact with the GUI straight from the Kernel without actually needing to "reserve" the UI and therefore enables parallel processing between the employee and the bot.

This drastically improve both response time and memory footprint, and ultimately the performance and resilience of your bots. Definitely very few RDA software can actually offer this. At the same time, agents deployed at the employee’s desktop level start gathering user behaviors in order to identify new automation opportunities. All of this without interrupting the ongoing work. The information is then used to identify every variant of any given processes. Slowly but surely, RDA scope starts expanding.

Optimized type (the next level of doing things)

Pretty much the same as “Advanced” expect for the automation discovery attribute. There is also an option to run some bots in a virtual desktop for additional load balancing (you don't expect however this to happen very often). At this level, RDA proposes or makes adaptive decisions, using information obtained from the workforce activities. RDA will constantly check and optimize the automation tasks. Adding self-monitoring and self-healing capabilities to RDA is a game changer.

If you want to read more about process automation, please check my previous posts or just?follow me .

Opinions expressed are solely my own and do not necessarily express the views or opinions of my employer.