Intelligent Automation, Not So Much Intelligent After All.

Intelligent Automation, or IA in short, combines Robotic Process Automation (RPA) with some forms of Artificial Intelligence technology.

Although Gartner predicts 90% of large organizations globally will have adopted RPA in some forms by 2022, there are many results gathered from different implementations of RPA showing the project did not meet the expectation mainly because of two reasons:

the automated process is more dynamic
the process is not robotics and is more on decision-making

The RPA vendors try to push the solution at every opportunity, but the nature and foundation of RPA, cannot deal with change and dynamic behavior. If a process is repeatable and static, that's the first criteria for where RPA is a good fit.

RPA vendors try to incorporate some forms of integration into AI/ML technologies to enable some use cases i.e. image recognition, NLP, Speech recognition and etc., with the hope of reinventing themselves for more dynamic and changing environments. And that's the market of Intelligent Automation today.

Nothing is intelligent about the current state of Intelligent Automation!

If we look at the core of both reasons RPA is failing, neither of the issues will be addressed by Intelligent Automation.

ML technology is not about decision-making. It is about identifying, predicting, classifying and etc., algorithms supporting decisions. Combining predictive and prescriptive models makes the foundation of decision-making.
Change cannot be addressed by the process thinking approach. What drives the change are business rules and business decisions.
Dynamic processes in regulated environments require managing change through the lens of laws, legislation, and internal and external policies – the business rules.

The type of decision-making that organizations need is not at the level of Artificial General Intelligence (AGI), at least today! They are all about tactical and operational business decisions. Therefore, we need to enable RPA bots to make those decisions.

There are two options to enable bots to make decisions: choosing an RPA vendor with decision-making capability such as decision robotics. Or integrate decision management into RPA projects.

Let's look at an example in a retail industry where they bring suppliers onboard on the day-to-day operation.

The process is to a person:

Inspect the registration of the company on the government's company registration portal.
Validate the information provided on the website to ensure supplier registration and provided information match and are valid.
Scoring the supplier based on different criteria, Determine the result of the assessment. I.e. Accept or Decline.
Send the decline or acceptance notification to the supplier.

In this process, step 1, can simply be automated by an RPA bot. However, steps 2 and 3 are decisions to be made. And step 4, is to carry out the actions based on the decision outcome.

And this is how Intelligent Automation should work, integrating automation of manual tasks with decision-making technology to make a decision on behalf of a human.

The benefit of this approach is the decision-making piece uses the right decision automation technology that can: