Intelligent-Automation-Reference-Architecture

Building an Intelligent Automation Reference Architecture

When automating business processes, it is easy to shorten design time and jump right into development when your initiative is built around a single tool or a small set of tools. There is a tendency to view that toolset as the solution to all your organization’s automation needs. This is especially true when the business or your center of excellence (CoE) feels it has a clear understanding of the process being automated. But, when automating your business-critical processes, quick wins with your incumbent toolset can sometimes be a Pyrrhic victory, exacting too great a cost to be worthwhile.  

A common pattern in automation is to eliminate a bottleneck in one area using the CoE’s toolset, only to see a new bottleneck arise in another. Backlogged transactions are moved downstream and cause issues with new parts of the organization that are not ready for the newly automated throughput. Those downstream teams can each have a different issue that requires different approaches and/or tools. This pattern can bog down even the most successful CoE as it works to solve individual issues.

Most organizations have hundreds of processes that could be automated, and the time needed to properly design each one can seem daunting. There are many good reasons to invest in design, but first and foremost is that it presents an opportunity to develop an intelligent automation reference architecture.

What Is an Intelligent Automation Reference Architecture?

A reference architecture is a set of documents and diagrams that provide a templated solution or pre-defined patterns for business process automation. It sets forth prescriptive guidance for architects and developers as they work to design automations for specific business processes.

 An intelligent automation reference architecture should include:

  • A decomposition of the business processes with required features for functional areas like mainframe automation or optical character recognition
  • Technology and component selections for each type of automation use case, including infrastructure recommendations
  • Technical descriptions and diagrams of how different components can and should interface
  • Instructions for how people should interface with those components

What Does the Automation CoE Have to Gain?

A reference architecture improves your automation quality and efficiency in several ways. Since it is continually modified by your CoE over time as it encounters more use cases, it improves resource allocation by shortening design phases. A reference architecture creates a focus on complexity during the design phase in which small proofs of concept are built to determine if a component is suitable for use in an intelligent automation. It then can guide product decisions and inform future design phases for similar business processes. And, because the strategies in the reference architecture reduce the uncertainty of automation design, it creates documented successes.

It also increases predictability by creating an ever-expanding body of knowledge that allows you to plan development and implementation more accurately. Experience in similar processes can provide a useful framework for estimations and assumptions. And it creates more insightful evaluations of toolsets based on greater experience and documentation of tool, product and platform selection in the context of your organization’s business processes.

How Do I Get Started?

Start by looking at the tools your CoE uses today and other intelligent automation tools in the market that may solve problems your team has put on the back burner. 

A diagram of your RPA tool architecture combined with your optical character recognition (OCR) product of choice may be the complete as-is reference architecture for your team. Look at projects that have been viewed as “nice to have” or others that have been added to IT’s development backlog. Maybe you have a business process that requires reading and understanding unstructured data in customer emails. In that case, begin looking at natural language processing (NLP) products and investigate how they can be used to read those customer emails and how they could be integrated with your current reference architecture for future solutions.

Start with known use cases the CoE has not yet pursued before you begin thinking about unknown use cases. A detailed investigation into machine learning might seem like an interesting project, but a reference architecture still needs to focus on the end goal of providing value to the business. Your reference architecture will evolve over time but will allow the CoE to identify a “best fit” technology for specific needs versus using the one-size-fits-all approach. Many valuable automation opportunities are out there; they require a multi-technology, holistic approach. 

ISG provides enterprises with an intelligent automation reference architecture and the experience needed to implement and harness intelligent automation platforms to fundamentally reshape the way they work. Contact us to find out how we can help you.

About the author

Cass Bishop is a Director in the ISG Automation Practice. He is a dedicated leader who implements global client-focused solutions which drive dramatic improvements to business process and automation lifecycles.  Cass brings over 20 years of Technical Implementation and Program Management experience in transformative technologies with a focus on System Development, IT Automation, Cloud, IT Service Management and Business Process Automation.