Automating and Educating Business Processes with RPA, AI and ML
How are these technologies connected, what are the implementation challenges, and how are companies using them?
Robotic process automation, artificial intelligence and machine learning are all being infused to automate business processes and speed time to decision. What is the "sweet spot" for each of these technologies, and how are companies using them? The common touch point for these technologies is automation.
When you use RPA, you are automating repetitive tasks, so staff doesn’t have to do them. An example is defining and implementing a robotic process automation process that automatically screen-scrapes invoice information from one system and enters it into another system, without an office staff having to manually key information from one system to another.
When you use AI, you are adding automation to decision making. Instead of performing a supply chain risk assessment manually, you enter a diversity of relevant data points into an AI data repository, and then present several what-if risk scenarios that you want the system to analyze and return answers for. The AI system comes back with several different potential outcomes for each risk scenario and then you make the final decision.
When you further augment AI with machine learning, you activate an AI system's ability to detect and analyze data patterns on its own, and to “learn” from those patterns. The advantage of this is the speed at which the system can process data and recognize patterns on its own that a human couldn’t. What the machine learning discovers has the potential to reduce your speed to insight of an important pattern or trend developing in the situation you are studying so you can respond to the situation sooner.
In summary, RPA automates routine, repetitive office tasks; AI adds automation to decision making; and ML is an ongoing educational process for the AI so the AI can “learn” from the patterns and trends developing in the data points that AI is charged to evaluate. Collectively, RPA, AI and ML all play important roles, and must be intelligently orchestrated as tools for business process automation and education to occur.
Overcoming implementation challenges
In working with cognitive automation tools, a major hurdle that many organizations face is understanding which tool to use when.
Here are four common challenges that enterprises face in their adoption of RPA, AI and ML:
1. Unrealistic expectations
In late 2017, a Deloitte survey on RPA revealed that 53% of enterprise respondents had already begun to implement or at least test the waters with RPA. This was a figure that Deloitte projected would grow to 72% of organizations by 2020.
According to Deloitte, most of these organizations were looking for continuous process improvement for their workflows, with automation as a secondary goal. Yet, when Deloitte asked these same organizations about how well they were able to leverage and scale their use of RPA to other areas in their companies, only 3% said they were succeeding in doing this.
The Deloitte report stated: “Many organizations, having started by treating RPA as an experiment, are now 'stuck' and are suffering from IT issues, process complexity, unrealistic expectations and a 'piloting' approach,” said Deloitte. “Maximizing the impact of RPA requires a committed shift in mind-set and approach from experimentation to transformation.…Given the relative immaturity of the automation market, it is taking time for large organizations in particular, to learn about and to adopt RPA at scale.”
The story doesn't change much for AI and ML. Many companies are still working through proofs of concept that characterize early stages of adoption. They are not yet at the stage where these technologies can be broadly leveraged for maximum business benefit throughout their companies.
One element slowing expansion is limited on-staff knowledge and experience with these technologies, and how the technologies can best be applied to business processes and decision making.
2. Education of executive management
Support for RPA, AI and ML from the C-level has been strong, but to assure long-term C-level support and budgetary investment, IT and data science departments must do two things:
They must produce successful implementations of these technologies that return tangible business benefits.
They must educate non-technical C-level management on the differences between RPA, AI and ML tools -- and how all of these tools come together in a business process or operation.
Upper management education is critical if the CEO and others are to feel comfortable going before their boards to explain and to field questions about these technologies, and why they are investing in them.
3. Vendor cooperation
I once directed an IT systems integration project in which my team had to work with several different vendors to implement the integration. Each vendor came with its own API and insisted that the other vendors use that API. It took us several weeks of negotiating with these different vendors until we could all agree on an integration approach. This took valuable time away from the technical project work. Integration complications like this equally apply to RPA, AI and ML.
Ease of integration matters because It is unlikely that every tool IT or users purchase from RPA, AI and ML vendors will be from the same vendor. Vendor cooperation will be needed when you want to integrate and scale solutions for your business.
For any RPA, AI or ML vendor you vet, the ability and willingness to cooperate with your own company and with other vendors should one of the first questions you ask about.
4. User engagement
RPA is the automation of a manual business process so that users no longer have to do it. It’s users who are in the best position to identify the repetitive processes that they would like to eliminate, and users who know how to define the business rules that the RPA must perform in order to successfully execute the process.
The same principle applies for determining the types of decision support needed from AI to support the business. What problem does the business want to solve? Without continuous user engagement, there is risk that IT/data science drifts from what users want. That can spell disappointment and even failure for a project.
Ensuring successful implementation of RPA, AI and ML
Successful implementation of RPA, AI and ML begins with understanding the differences between these automation tools and how they are used -- and mastering the way in which they are applied to the business cases your organization needs to address.
There are organizations that are doing this and getting impactful results.
Chinese e-tailer Alibaba Group uses AI to help map the most efficient delivery routes for merchandise, and U.S. e-tailer Amazon uses AI to predict product demand and to tailor product recommendations to customers.
Medecision developed an AI algorithm that was able to identify eight variables to predict avoidable hospitalizations in diabetes patients.
The UK used RPA to automate the issuance of citizen reminder letters.
“We believe that every large company should be exploring cognitive technologies,” stated Thomas H. Davenport and Rajeev Ronanki in the Harvard Business Review. “There will be some bumps in the road, and there is no room for complacency on issues of workforce displacement and the ethics of smart machines. But with the right planning and development, cognitive technology could usher in a golden age of productivity, work satisfaction, and prosperity.”
Davenport and Ronanki are right. The potential is there, as are the technology “wins” for companies that adeptly target business and decision processes that will benefit from cognitive automation.
Learn more about AI, RPA and ML in these articles:
Enterprise Guide to Robotic Process Automation
About the Author
You May Also Like
2024 InformationWeek US IT Salary Report
Aug 15, 20242024 InformationWeek US IT Salary Report
May 29, 20242022 State of ITOps and SecOps
Jun 21, 2022