The 8 dimensions of intelligent automation

Robotic Process Automation(RPA) , when combined with a Cognitive Technology like machine learning , is a powerful tool for automating a process and achieving the maximum possible level of automation you can apply to a process. But should you apply this to each and every process you see in your operations landscape? The answer depends on multiple factors and if often no. There are multiple factors that you should consider , some of them inputs from your business and operations teams, and some of them from your technology teams. They can be categorized into 8 areas within two categories:

Business and Operations Inputs:

1. Process Volumetrics

   Factors like the volume and frequency of a process are important considerations for Robotics Suitability. As a rule, process which have volume and frequency should be prioritized for RPA as potential business benefits are the highest in such cases. Process with sufficient volume and high error rates should also be high on the priority list if the potential for error reduction is high post robotic automation. Another factor that should be considered is the possibility for a process to run for 24 hours post automation when it is only being manually executed during business hours before as the ability for a software bot to work round the clock is a key benefit of RPA.

2. Process Inputs and Outputs

   The kinds of process inputs are key to determining suitability. The ideal candidate for RPA is a process that is initiated from structured input sources like an excel sheet or a webpage with a consistent format. Printed Forms can also be considered as structured sources as an OCR engine would be required to pull information from such a source and feed the data to downstream systems. Semi-structured data from invoices and unstructured data sources like emails require the use of a cognitive engine to extract data and will be more complex.Also, the number of input formats should be considered. A process may have voice, emails , web forms and other data sources and the larger the number of input sources, the less suitable it is.Also, the variability of input within a particular type of input should also be considered. Similar rules apply for output formats.

3. Level of Process Standardization

   The level of process standardization is an important, and often ignored factor in process suitability frameworks. It is important to consider the cycle time variations, the consistency, and the presence of wasteful steps in a process. Applying RPA to an inefficient process often makes it more inefficient.

4. Business Benefits

   The impact of automating a process is much more than the potential FTE reduction . RPA can lead to business benefits in customer facing processes when the turnaround time is much faster, better compliance with SLAs and reduced risk for clients and counterparties. Reduction in error rates can also reduce fines and remove wasteful effort devoted to fix errors. There are also softer benefits like better employee morale which business cases do not usually consider. Processes are sometime part of existing automation initiatives and this is sometimes missed when a proposal is made to apply robotics to a process

5. Effort to Automate

   The complexity of a process if strongly proportional to the time taken to automate it. Complexity can be thought of as the number of decision points in a process and this determines the effort to apply rules and automate it. Highly complex processes ,which can't be modeled using a set of rules ,are candidates for Cognitive Automation.

Technology Inputs:

1. Systems and Applications Involved:

   The list of systems and applications that are used by operators is a critical factor in determining RPA suitability. Applications accessed with virtual environments like Citrix are traditionally more difficult to apply RPA to. RPA software usually support a default browser like IE and the availability of an application via IE should be considered. The applications involved should be accessible on the same system

2. Cognitive Elements in a Process

   Unstructured data and fuzzy rules in a process make it more difficult to apply rule based RPA to a process. For example Emails and Handwritten content require the availability of a cognitive engine to enable end to end automation. Process that involve reconciliation also require some machine learning.

3. Technology Availability

   A process might be suitable for automation with the above criteria but the technology might still not be deployable in the firm . Factors like the availability of Robotics tool licenses , the open source software policy, the budget and the review processes to bring in a new cognitive tool are factors which might make a lot of analysis irrelevant in the end.