The Impact of Cloud, AI/ML and RPA and on Decision Management

Decision making is a key component of today's business applications. Complex business applications must be able to make decisions following the same rules a human would follow when making those same decisions.

Because business applications are so critical –  and because they need to incorporate business know – how into the applications to support accurate decision making – the building of these applications is no longer left solely to IT developers. Software development is seeing increasing involvement from the business side. 

For example, in the past an insurance company would write an application that records insurance claims. Today, IT is writing applications to sell insurance. That is a huge change. In computer science class, developers are not taught how to sell insurance. And it is not like the insurance company lacks that expertise anyway. They have many people who know how to sell insurance, but none of them work for IT. So more businesses are realizing that they need to involve the business stakeholders in the software development process, and incorporate more business knowledge directly into these applications. 

Clearly, the business stakeholder are not developers. They are not going to write code, but they can produce models of their business, based on the business rules, the processes, the policies, and the decisions they make while conducting company business. These models can be thought of as the source code that will be deployed within the business apps. 

How does IT empower the business user to encode as much of their knowledge as possible so they can add value to the applications the organization depends on? Development teams should utilize business rules rather than simply encoding business logic into applications. When business logic is built into the application – which is the traditional way of developing business applications – the business stakeholders have no visibility. It is hard for them to see what specific policies are being applied; what the rules are; and why a decision is made. 

The solution is to outsource the decision making to a business rules engine, which makes decisions on behalf of the application, based on a set of rules written in English so the business side can understand. Business can gain much more control over the policies that are applied automatically by these applications.

A business rules engine provides multiple advantages, including:

  • Separation of the business rules from the applications
  • Visibility for business stakeholders
  • Business rules expressed in terms that the business can really understand
  • Enabling business and IT experts to collaborate more effectively
  • The ability to change rules easily and quickly
  • Consistency of rules and decision-making

A range of Red Hat customers across various industries are having great success using a rules engine to provide decision services to the organization and keeping business rules separate from application code.

A lot has changed in IT over the past few years that has impacted automated business decision making. In this blog, I will cover three of the major changes that have caused the greatest impact: Cloud, Artificial Intelligence (AI)/Machine Learning (ML), and Robotic Process Automation (RPA).

The Cloud

The cloud has dramatically altered the way we create, deploy, monitor and manage business applications, and that has a tremendous impact on people using business rules engines.

Since the start of IT, organizations have built “Monolithic” applications, which are large, difficult to understand, and challenging and time-consuming to make a change. In the last few years, thanks to the cloud, monolithic apps have been replaced by containers, and microservices architecture, making it easier to create, deploy, manage and change parts of applications. A large app can be broken into smaller components that can be developed, scaled, managed and changed independently from the other components.

The flexibility of containers and microservices simplifies decision-making because rules can be added as components rather than being embedded in the application. In the cloud, rules can modified easily in response to changes in the market or in the way the company does business, and the entire application does not have to be rewritten. This provides businesses with a level of agility they never had before.

Artificial Intelligence (AI)/Machine Learning (ML)

Another set of emerging technologies that is having a major impact on business applications is Artificial Intelligence (AI) and Machine Learning (ML). 

One way of defining ML is “rules that write themselves.” This is obviously a huge leap from where we were 10 years ago. Instead of creating the rules based on your experience and building an app based on those rules, you can now look at historical data, figure out what rules produced those results, and then apply that knowledge to make decisions going forward.

Essentially ML is used for constructing predictive models. For example, with regard to the rules covering when insurance claims are denied or paid, you have data about how you arrived at decisions in particular cases. You could use ML to build predictive models to repeat those decisions in similar cases. 

How do predictive models compare with user written rules in terms of their usefulness and viability for these types of decision applications

  • Rules are created by people, while predictive models are automatically created based on analysis of historical data
  • Rules produce results that are explainable, while predictive models produce results that are not explainable.
  • Rules are subject to human error. One of the challenges with rule-based systems is you get gaps – such as situations you forgot about, or edge cases you did not address. Conversely, predictive models are subject to historical bias. You are limited to reproducing behavior apparent from the training data. So if the data is biased, you get a biased model. 
  • Rules take a significant amount of time to produce, while it is relatively quick to generate a predictive model once you have the data.

The goal when using ML to augment decision making is to combine the advantages of applied rules and predictive models. New rule languages, like Decision Model and Notation (DMN), make this possible by simplifying the process of creating rules.

In the past, rules were created in a complex notation. Thousands of rules leave open many possibilities for error. DMN is a graphical language for encoding rules that make a decision. It makes it much easier for business stakeholders to create the source code for their decision applications. They create a graph to encode complex business logic. Then they can incorporate a predictive model into the DMN diagram, and they can incorporate business rules as well, effectively combing the best of both options

Robotic Process Automation

Robotic Process Automation (RPA) is an exciting, fast moving space right now. With RPA, software robots are developed to perform routine, repetitive work that would otherwise be done by a human worker. The advantages of RPA are all about reducing cost and headcount by automating tasks.

Time spent copying information from a back office database into a spreadsheet, and moving data around from system to system – this is not the type of work where a human worker adds value to the process, but it still has to be done. RPA allows an organization to automate those repetitive tasks. 

One of the best advantages of RPA is that it enables you to automate work without having to change your underlying systems. You can keep your legacy mainframe and other applications exactly as they are. The robot pretends to be a human and carries out the same tasks exactly the same way a human worker would. 

But it is very important to consider RPA as just another software development approach. In future, as RPA evolves, I would expect to see containerized robots roaming around your hybrid cloud. But for now, RPA is just another app and it needs to be managed just like any other app. Version control and QA are very important. 

When you create a bot, beware of the attack surface. You create an opportunity for someone to add a few lines of script to that bot. Think about the damage that could be inflicted by an automated bot with high level access to all your enterprise data. That is a key reason why RPA should be subject to the same governance as any other software.

Also it is very important to understand that the value of RPA is in the ability to automate human work, not to patch holes in IT systems. If you find yourself building bots to fix holes in IT, you really need to take a good look at the infrastructure instead. It is not productive to use bots as band aids, because the bots themselves will continuously break as things change within the infrastructure. It is more productive to focus on fixing the source of the issue in the underlying infrastructure.

For RPA to remain relevant and continue to support software development, bots should be compatible with the cloud, and be able to run in containerized environments. This is technology we expect to see in the next few years or so.

The Red Hat Solution

Red Hat is very active in the software development space and offers a range of tools designed to solve the challenges associated with incorporating rules and decision making into business applications:

  • Red Hat Decision Manager is a platform for developing containerized microservices and applications that automate business decisions. Decision Manager includes business rules management, complex event processing, and support for building DMN models.
  • Red Hat Process Automation Manager is a platform for developing containerized microservices and applications that automate both business decisions and processes. Process Automation Manager includes business process management (BPM), business rules management (BRM), and business resource optimization and complex event processing (CEP) technologies. It also includes a user experience platform to create engaging user interfaces for process and decision services with minimal coding. 
  • For development in the Cloud, Red Hat OpenShift is an enterprise-ready Kubernetes container platform with full-stack automated operations to manage hybrid cloud and multicloud deployments 
  • Red Hat Runtimes is a set of products, tools, and components for developing and maintaining cloud-native applications. It offers lightweight runtimes and frameworks for highly-distributed cloud architectures, such as microservices.