When thinking about your automation journey, it may seem like you are on the way to Oz via the yellow brick road. Lions and tigers and bears are lurking in the forest to trip you up or send you off the beaten path. There is no question, dangers are lurking in the automation market. One of them is just the sheer volume of tools and the confusion created by such rapid change. Attended versus unattended, artificial intelligence, machine learning – what does it all mean?
When harnessing the power of automation, an organization knits together a set of technologies to bring about true digital transformation. Of course, it may make sense to commit to just one automation technology to start with, but to optimize the power of automation and identify where these technologies can best be applied, an understanding of all the available options is important.
Here’s an automation technology primer:
- Robotic process automation (RPA): RPA uses the graphical-user interface or (GUI) of a client system to automate high volume, structured and repetitive processes. RPA mirrors human actions by following structured business rules executed by humans today. In cases where RPA can be employed to replace human labor, teams are freed up to do more. RPA is best considered the “hands” of automation. It executes predetermined actions in precisely the same way each time. Because it always achieves the correct outcome, it increases confidence in compliance.
RPA can be attended or unattended. An attended automation is one that directly interacts with a human to provide real-time support for a human-executed process. For example, an attended automation may be used by a call center agent to gather data across disparate systems while on the call with the end customer, reinforce the process steps needed for a certain type of end-customer interaction and improve compliance with required, scripted customer disclosures. An attended automation can run locally or virtually but is not run in a centralized control room like an unattended automation. Attended automation opportunities can scale in a similar manner as unattended automations by taking on common tasks or processing steps across a population of resources doing the same thing. An unattended automation typically runs on a pre-defined schedule without human intervention, like batch processing. Unattended automation is where many organizations start their automation journey, and many believe they “max out” the opportunities. But when considered together and in conjunction with other technologies, unattended and attended RPA solutions can be applied in so many ways.
- Image recognition and ocular/intelligent character recognition (OCR/ICR): OCR is the next stop on the yellow brick road and the next rung on the automation complexity scale. This technology is used to “read” text – primarily digital characters with some capability to read handwritten notes in certain languages – and transcribe it into various systems. OCR is often used in conjunction with RPA by digitizing and standardizing input data to be consumed by RPA. Image recognition is automating the process of identifying and detecting an object or a feature in a digital image or video. Depending on the use case or applicable business scenario, one or more OCR/ICR tools may be needed.
- Natural language processing (NLP): NLP is the application of computational techniques to the analysis and synthesis of natural language and speech. NLP is used to understand the intent in an email or service request in customer relationship management (CRM) and other systems. NLP takes unstructured text and provides a structure for RPA to follow through and process. Though it can be complicated to implement, once set up, it operates autonomously and is self-learning. The more data that can be fed into an NLP engine, the better the system will improve with every process run.
- Cognitive (machine) reasoning: Cognitive reasoning technology makes complex decisions based on previous knowledge and provides rationale as to the decision made. By using knowledge of a process to make a more complex decision – including why that decision was made – cognitive reasoning is designed to augment the human workforce in making better, more data-driven decisions. They key to cognitive reasoning is good data.
- Conversational artificial intelligence (AI): Conversational AI is the evolution of NLP that underpins a virtual agent that can hold well-structured conversations. Anywhere a conversation takes place, such as common banking transactions, FAQs, or IT service desk activities, is an opportunity for this technology, commonly known as a chatbot.
Potential pitfalls can certainly waylay the automation journey. The key to making it down the yellow brick road is to ensure your automation initiative is an enterprise program and becomes foundational to your ways of working – not a “fly by night” affair.
A few of the most common pitfalls and strategies to address are:
- Change management: It is critical that the enterprise embrace the automation initiative. It should be conveyed as a transformative program that is moving the enterprise forward.Celebrate benefits such as productivity improvements, compliance and revenue generation. Lack of effective change management is one of the reasons why enterprises hit the “RPA wall,” where automation programs often stall.
- Methodology and governance: A framework for delivering automation is imperative in any organization. Many enterprises new to automation think they know – but often don’t – how to manage process detail for automation, product selection, contracting, implementation best practices, center of excellence and operating model standards, tools and techniques. An established automation lifecycle and governance framework can help keep automation initiatives on track.
- Enablement: This goes hand in hand with change management. Automation is not a “one man show.” It is a team effort that must be orchestrated between IT and the business. Enablement of the entire automation team, citizen developers – if you choose to use them – and the stakeholder community is necessary for success.
- Benefits realization: What happens when automations are developed and running but benefit realization is minimized or inconspicuous? This often happens when the benefits are not validated or committed before starting. Benefit calculation and confirmation should occur before and after developing your automation. Consider using standardized estimates for development and licensing costs and align on how you define savings in the business case such as cash savings and non-cash savings.
How can you ensure your trip to Oz is safe? One of the most important principles to consider is a broadening of your approach – imagine various technologies working together harmoniously with humans to deliver real business value. When done right, automation can enable a process from end to end. For example, RPA ingests various inputs and parses them into categories for processing. One category may require OCR specialized for handwriting, one may require NLP. Once the processing is complete in the respective technology, the RPA spits out a summary of results and performs updates to the business system or application.
Automation is not a fly-by-night activity. It is a transformation to the future way of working! Talk to others in your industry and ask how they have achieved automation success. Even if you’ve hit a plateau, don’t fret. You still reach Oz safely.
ISG helps companies navigate the complex automation market, identify process candidates that promise healthy return on investment and scale capabilities across the organization. Contact us to discuss how we can help you.