Reshaping business processes by using AI is the future

In March 2023, Holweg, Jeavons, and Davenport released an article about A.I. and its possibility to help companies reshape their processes. Business process reengineering was extremely popular in the 1990s. Businesses employed cutting-edge tools like enterprise resource planning (ERP) systems and the internet to implement drastic modifications to extensive, end-to-end business processes. Companies expected revolutionary improvements to broad processes like order-to-cash and conception to commercialization of new goods as a result of reengineering, which was supported by academic and consultancy proponents.

Nonetheless, despite the fact that technology did bring forth significant upgrades, implementations frequently fell short of unrealistic expectations. Large-scale ERP systems, such as SAP or Oracle, for instance, generated highly inflexible procedures that were difficult to adjust after the IT adoption while simultaneously providing a helpful IT backbone for data sharing. Since then, process management has often just required small changes to local processes, all done without the use of technology. Examples of this include the use of Lean and Six Sigma for repetitive activities and Agile Lean Startup methodologies for development.

At certain businesses right now, a variation of this concept is returning, and we anticipate seeing it in more. It will necessitate not only a respect for and knowledge of AI but also a fresh perspective of business processes as a framework for enhancing work. The radical reengineering of business processes that the initial proponents of reengineering had in mind seems more and more likely to be attainable as AI develops into a generally applicable, general-purpose technology. (One of us, Davenport, authored the initial book on the subject.)

Current Reengineering

In the 1990s, transactional and communications-based technologies enabled reengineering in large part. They made it possible for effective data gathering and transfer both inside and across organizations. On the other side, AI allows more automatic, quicker, and smarter judgments. Essentially, the majority of AI implementations in big businesses entail learning from massive datasets to generate a forecast or a classification, which then aids business in making better operational decisions. By resulting in better outcomes, smarter operational decisions also increase efficiency. The fact that current AI systems are truly general-purpose technologies and have significantly altered not just production planning and control but also visual image identification and inspection, autonomous operations, and the creation of new content is a crucial distinction.

The cost of adopting the techniques that are driving the development of AI has dramatically decreased even though they have been present for many years. Modern AI-based solutions, which were previously solely available to data scientists, are now sufficiently developed to be sold "off the shelf," significantly reducing the technical hurdles to entry. Model-driven prediction is now much more affordable thanks to falling processing costs, which are being fueled by the widespread use of the cloud, the expansion of low-cost internet, and decreased sensor prices. The broader definition of automation can also incorporate AI-based judgments. Automation of information-intensive back-office tasks is made possible by technologies like robotic process automation (RPA). RPA can to a limited extent use data-based judgments because it is rule-based. It can handle a far wider variety of jobs when integrated with machine learning as "intelligent process automation," though.

This AI-driven reengineering is now underway. It is being used by banks to change the way they advise clients on wealth management. As well as automating claims estimates for vehicle and house damage with deep learning analysis of images taken by the insured, insurance firms are employing AI to make customer onboarding and underwriting more simpler. Engineering and maintenance procedures are being altered by industrial businesses. Diagnoses and treatments are being changed in some nations by AI-based telemedicine, despite the fact that there has been a lot of study on AI but considerably less clinical use.

The way we utilize AI, how labor is done, and how businesses are set up are all affected by this in significant ways. Companies must reconsider their operations from the perspective of an end-to-end process and carefully consider how AI may alter them if they are to reap the benefits of these prospective advantages. Companies essentially need to investigate the areas where they are producing enough data to identify patterns that may be utilized to support operational choices.

AI Drives the Reengineering of Processes

Businesses must reconsider the activities that are required, how frequently they occur, and who does them as AI adds new capabilities to business processes. Businesses must also select what tasks people and machines will do in their operations when AI is complemented by some automation. The majority of AI programs to date aim to make a certain task better. The bigger picture, however, is being overlooked; astute businesses see the use of AI as justification for a fresh look at end-to-end procedures.

Process analysis frequently contains both limitations and possibilities at its most fundamental level. For instance, the manager of transaction surveillance (anti-money laundering and fraud detection) at DBS Bank in Singapore expressed frustration in an interview on the high numbers of false positives found by the rule-based system demanded by banking authorities. He saw a chance to apply AI to anticipate and rate the fraud risk of each good outcome using machine learning, despite the fact that this process limitation is inescapable. Transactions with low fraud probability might simply be "cold-stored" for a few months to determine if the same consumer was involved. In the field of fraud detection, AI systems that use machine learning to identify outliers are well-established. The productivity of surveillance analysts, however, improved by a third when the machine learning system was coupled with a fresh workflow platform and a method to identify fraud network participants.

Another excellent illustration is Shell, where one of us, Jeavons, oversees AI activities. The corporation Shell has historically been process-oriented, and it is currently working on a significant AI program in areas like supply chain, operations, and maintenance. Shell is reengineering its business procedures as part of this.

As an illustration, think of the monitoring and inspection tasks at chemical and energy plants, pipelines, offshore infrastructure, and wind and solar farms. Inspectors and maintenance personnel used to complete this task solely on site, but AI can reduce that restriction. Several low-value inspection activities may now be completed remotely by robots and drones. Drones and robots are being used to automate these operations and assist cut the cycle time since certain Shell installations are so big that it would have previously taken years to examine everything by hand.

Inspectors and maintenance specialists might now reconsider their regular job as a result of these adjustments. If they are on site, they may complete more complex verification while concentrating on higher-value tasks like prioritizing projects. The management of the training procedures for the tens of thousands of machine learning models that are currently in use is one of the new activities that are developing at the same time, as is the annotation of pictures to enhance inspection algorithms. Multidisciplinary teams working on activities that are primarily digital now oversee what were formerly physical job processes.

There was some pushback against this change. The inspectors were first difficult to persuade, but over time when they are demonstrated that image processing offers equivalent accuracy in a lot less time, they are won over. Shell is also helping these engineers rethink their work practices with the remote surveillance centers, giving them the freedom to lead the transformation.

Shell has discovered that this process of AI-enabled reengineering is becoming into a consistent manner of doing business. Each project may only take a year or two, but the more they utilize data, AI, and technology to reinvent procedures, the more potential they see for progress. This is crucial since the business is changing into a net-zero emissions energy firm.

Who Should Drive a Process Transformation Enabled by AI?

The exclusive purview of operations managers has always been process improvement. Because of this, it has been a little bit unusual for businesses to run explicit reengineering initiatives in tandem with their AI efforts. The AI project should include process design and improvement initiatives in order to fully capitalize on the capabilities of AI. The most effective of these programs are now being led by "product managers," whose goal is to successfully deploy the technology while implementing the necessary business changes. Shell appoints a product owner to oversee the business transformation and a product manager to oversee technical execution. Some companies also practice "design thinking," which has some overlap with reengineering-style studies of how processes and activities should be altered to fulfill external or internal requirements.

Although we have observed several cases where reengineering occurs along with AI development, not enough firms have yet to acknowledge the need for process transformation. Whether it is labeled "reengineering" or not, a clearer definition of the reengineering function and tasks, such as high-level design, specific process flows, assessment of before and after costs and cycle times, and analysis of required skills and training, would be beneficial. To leave the tasks to chance or a smart manager who is familiar with the reengineering movement would be a grave mistake given how crucial they are to the success of AI initiatives.

Automation-focused initiatives are more likely to have a formalized set of process improvement processes since they directly affect process flows and are more likely than other types of AI to simply entail incremental change. For instance, Voya Financial maintains an automation center of excellence within its process improvement division, and no automation project is undertaken without first attempting to enhance the existing processes. The group's leader informed us that within the organization, automation is just as much a process-oriented engagement as it is a technological one. Nevertheless, we'd want to see a mix of more aggressive process transformation and more potent AI technologies like machine learning. We've seen a number of other businesses that combine automation with process improvement.

AI is quickly spreading across all industries. It will become as commonplace as ERP systems, statistical software, or even spreadsheets once the buzz dies down. A far wider range of businesses may reengineer their processes by using AI platforms. The use of AI is a tool, not a goal in and of itself. The long-term benefits of AI are likely to benefit businesses that can use it as a new tool within the larger framework of process reengineering.