Intelligent Automation Solutions in Finance

Author: Bill VonMinden

What is Intelligent Automation (IA) and what can it do for my organization? 

Intelligent Automation harnesses the power of technologies such as Artificial Intelligence (AI), Machine Learning (ML), and Robotic Process Automation (RPA) to save time and money. It increases efficiency and scalability while reducing costs and errors. And by enhancing decision-making capabilities, driving operational effectiveness, and highlighting candidates for process optimization, it has the power to transform your business. 

Intelligent Automation is a blanket term that covers several closely related technologies which form an ecosystem supporting the identification, automation, monitoring, and optimization of business processes. A few years ago, Robotic Process Automation was the most common technology in the space. However, technologies have evolved and matured, and now Intelligent Automation is the preferred term that refers to all the technical components working together. 

In this article, we will break down the different components of IA and show how they can be applied to a simple use case in finance. 

Components of Intelligent Automation 

Below is a list of some technologies that fall under the Intelligent Automation umbrella and their benefits. Some of these technologies help companies identify inefficient processes, redesign those processes, and then monitor them, while others either automate manual processes or inject additional intelligence into processes that are already automated. 

Process Mining 

Process Mining enables organizations to easily identify candidate processes for automation opportunities. Today, most systems support various levels of logging. These logs provide a breadcrumb trail of activity generated by users’ day-to-day operations.  

Process Mining tools analyze this log data to find patterns of usage that indicate a repetitive business process that could benefit from automation. Other process mining tools may involve software agents monitoring the activity of the user, directly on the workstation. And, of course, there’s always the manual approach of having a human observer monitor the users directly. 

Process Monitoring  

Using the same logs mentioned above, we can begin gathering data on the process over time, both establishing a baseline for performance and tracking changes from that baseline over time. Monitoring and measuring the process lifecycle provides insights into bottlenecks that create unnecessary expense, delay, and opportunity cost.  

In some cases, processes must be closely tracked because they support critical functions involving regulatory compliance or other risks to the business. Process monitoring allows us to quickly document the return on our investment as well as identify additional optimization. 

Process Optimization 

Once a process has been automated and can be measured against established KPIs, the rich logging infrastructure provided by the automation can be used to guide further improvements, highlighting inefficiencies, and providing a basis to re-envision the process. After automation, historical data is processed to show which parts of the process introduce unreasonable cost, delay, and risk, providing further suggestions to optimize the process. 

Robotic Process Automation (RPA) 

RPA is what people typically think of when automation is mentioned. RPA “bots” provide the ability to quickly and efficiently operate nearly any software that a human can use. RPA provides the bridge between components, tying them together to automate processes and deliver value to your organization. In one form or another, RPA has been successfully used for more than a decade, and today’s bots leverage advanced technologies to operate a large swath of applications, using little or no code.  

Artificial Intelligence 

Automation “bots” are much smarter and more flexible than they used to be. They can push and pull data to or from other systems, allowing the user to see a more complete picture. They support collaboration between users. They are also increasingly being connected to AI solutions to further increase efficiency. 

Examples range from the simple, such as suggesting updates to wording based on generative AI tools or surfacing policies contained in a Large Language Model (LLM), to very complex, like recommending a decision on a loan application based on an applicant’s data and the organization’s rules and policies. Insights generated from data science analysis can also be incorporated.  

This integration of advanced tools and engines is what separates intelligent automation solutions from the simple “bots” of the past. 

Putting Intelligent Automation to Work 

To understand how each of the components listed above fits into an intelligent automation solution, let’s take a scenario where there are three steps in an expense approval process. 

Step 1:  An employee submits each expense along with a receipt.  

Step 2: The expenses are then either approved or denied by the function head.  

Step 3: Approved expenses are forwarded as a batch to the CFO, who is responsible for the final approval or denial. Approval notes can be added to each batch. Likewise, reasons for denial are sent back to the employee and function head. Approved expenses move forward for payment.  

Let’s say the company uses a custom, web-based expense management application in which all approvals are extracted into Excel and imported into the corporate accounting system. The expense management application tracks detailed usage information in its audit and system logs. That information includes the user that logged in and the details and disposition of each expense item.  

Even though the expense management system is built and maintained internally, IT resources are always in demand. When requested by the finance department, it can take the IT team several months to make changes to the system. This presents obstacles for the growing business, as it needs to adapt to new challenges quickly. 

In our scenario, we see a rich number of data sources that capture activities of the expense submission process. However, this also raises several questions: 

• How many approvals and denials are being issued each month? 
• How long does it take to approve expenses and resolve denials? 
• What kind of limits are in place for approvals/denials? 
• Where are delays happening in this process? 

All these questions can be answered, but not necessarily easily since the answers reside across two systems, one of which is custom. 

Applying the Tools 

Each of the tools discussed above can be applied to our scenario. 

Process Mining tools can be easily pointed toward the log files, highlighting the amount of time each step in the process takes and identifying the approval steps as processes ripe for automation.  Once the candidates for optimization have been identified, we can immediately start using Process Monitoring, building a performance baseline. At that point, we can begin optimizing the process, looking for steps that can be eliminated or ways that we can increase efficiency. 

The approval of expenses at both the function head and the CFO level are obvious candidates for automation. An RPA “bot” can be created to assist with approving expenses. Simple rules can be applied to automatically approve certain expenses.  

For example, cell phone expenses under $50.00 might be automatically approved. Expenses over certain thresholds or without receipts can be flagged for extra scrutiny, as can expenses for accounts the employee is officially responsible for. These simple rules catch obvious errors and improve efficiency, allowing the function heads and CFO to focus their time on expenses that require a closer look. 

Beyond these simple rules, AI tools can be incorporated into the process to provide additional value. These tools might look for unusually high or low expenses. They might use image recognition to compare the amount on a picture of a receipt with the amount of the expense item. Again, each of these examples not only reduces the amount of time spent approving expenses, but they also improve the overall quality of the process, catching errors that humans are prone to miss. 

After implementing the bot, we could regularly check in on the health of the process using process monitoring (in addition to actively managing our KPIs), quantifying the improvements in efficiency and reductions in time. We would also look for further refinement with the Process Optimization tools that could be implemented in another iteration. 

The Benefits of Intelligent Automation 

As you can see, Intelligent Automation is not just about basic “bots” anymore; it leverages simple checks within the bot as well as all the capabilities of AI tools to continuously automate and deliver value. Data quality is improved while employees spend less time on repetitive tasks, freeing them up to do other, more creative, more valuable work. Additionally, we can quickly identify the value in terms of increased efficiency, allowing us to easily quantify the return on our investment thanks to constant monitoring. 

If you are looking to increase efficiency in your organization, reach out to one of our experts to schedule a quick chat. We’d love to see if an intelligent automation solution can save you money, reduce errors, and make your employees happier. 

Bill VonMinden is an Architect at RevGen. In his four years here, he has worked on a number of successful projects covering Data Warehousing, Reporting, Integration and Intelligent Automation projects spanning the Healthcare, Telecom and Financial Services Industries.