Your paper lays out the three stages of the current digital transformation of the SCM industry, with a focus on the future state of incorporating intelligent workflows. What are those three stages?

We recently conducted a focus group with 15 industry partners and interviewed an additional six individuals who are actively involved in supply chain digital transformation efforts. Our investigation focused on how these organizations deploy and utilize technology – including AI – to achieve their transformation goals and how they are integrating both technology and human resources to address supply chain challenges. One thing we found was that the adoption of digital technology in SCM follows three phases: digitization, digitalization and digital transformation.

Digitization, sometimes called digital encapsulation, is converting existing data and documents into a digital format to accurately represent the physical world. In this stage, data is not altered nor analyzed, it is merely encoded.

Digitalization, on the other hand, entails altering processes, organizational structures and decision-making architectures to begin moving towards integrating processes across the supply chain.

Finally, digital transformation fundamentally changes the process to fully leverage these new digital technologies within and across key companies and entities, ultimately affecting how an organization creates value across its supply network.

With so many players having to buy into the value of these three stages and take the initial steps required to embark upon this digital transformation, where does the industry stand in working through these demands?

One of the challenges we face as an industry is how to ensure that all participants in a given supply chain are sufficiently prepared technology-wise to contribute to and help realize the benefits of the digital transformation being driven by major players.

One thing we found was that the vast majority of companies involved in our study were in the midst of a digital transformation effort but have not yet entered the final stage. Most had already taken the first step of digitizing their data, seeing it as essential to company survival. As one participant said: “Digitization of the supply chain is a requirement for being able to be an omnichannel retailer in the future. You’ve got to know what you have. You’ve got to know where it is with a high degree of accuracy. Or you’re dead.”

Many of our participating companies have also moved into the digitalization phase, using digital technologies to augment business processes but not fundamentally transforming them – i.e., providing their human managers with dashboards, inventory trackers, alerting systems, and even robotic process automation bots to improve the efficiency and effectiveness of their processes. Human managers, however, are still central to the execution of the process.

What do these companies need to do to move beyond the adoption of digital technology into what you call the “digital transformation” phase?

That’s where intelligent workflows come in – AI can only do so much. There are still cases in which AI models’ predictions are not correct and humans are needed to detect when the conditions under which the model was trained have changed. While AI will only get better, it will never be able to do everything.

An analogy might be helpful here. I was in the Air Force, and we used a lot of AI even back then. I flew a lot, and we’d have these big Airborne Warning and Control System planes (AWACS) in the sky letting us know in real time what was going on. We would get a briefing before we took off, but once we got into the air, we were updated on the contingencies of the moment and would act upon that real-time data as it came in. When we came back from our mission, we would debrief our experiences and all that new data would go back into the system for the next time we flew.

It became an iterative process whereby the system would propose a set of actions and counteractions based, in part, on how well previous recommendations worked out or didn’t, offering a more robust set of options to consider going forward. That’s how intelligent workflows operate in supply chain management.

This seems to point to a more strategic role for humans in optimizing the performance of supply chains based on a constantly expanding and increasingly detailed set of data and decision points, not a lesser one, right?

Exactly. The orchestration of intelligence – “artificial” and human – is what intelligent workflows are all about. An intelligent workflow implementation plan would provide parameters for the integration of all interactions among digital workers, human workers, AI and other IT systems necessary to complete an efficient, high-performance end-to-end business process.

In this scenario, Robotic Process Automation (RPA) uses software bots to automate repetitive tasks, freeing up human time for more complex work while AI handles many cognitive tasks such as first-level decision-making, natural language interaction and content capture. Finally, human experts support the overall intelligent workflow through ongoing quality assurance – handling cases that automation and AI cannot manage and investigating/resolving issues where digital technologies have been unsuccessful. Human experts would also provide feedback into the intelligent work to improve process automation and AI processes for retraining and continual improvement of the intelligent workflow.

In this sense, the human worker becomes a supporting actor concerning the workflow itself. Though their skills are crucial to the ongoing success and improvement of the workflow, humans are not directly responsible for working through the end-to-end business process. That responsibility is assumed by the workflow orchestration service, the intelligent workflow.

What does that leave for humans to do in adding value to supply chains in an intelligent workflow environment?

Subject matter experts and other SCM personnel add unique value to an intelligent workflow-based supply chain in at least two key ways: overseeing risk management and ensuring operational resilience.

The ability to respond to contingencies on the fly is a critical human function that AI can’t do on its own – there are too many intricacies, too many contingencies, factors being created the more you build out the network with all the accumulating data. Correcting the AI-based intelligent workflow model when it fails and providing the feedback necessary to make sure that same mistake isn’t made again is a crucial element of mitigating risk across the supply chain. That’s one essentially human role.

Another is instilling and ensuring a level of system-wide resiliency across the supply chain itself, the ability to not only address changes as they occur but also identify and plan for those likely to come. That goes beyond merely bouncing back from a disruption, for example, but transforming the process itself – the interaction of the technology and the people. AI systems in and of themselves aren’t that good at predicting outcomes – people remain better equipped for that important role, provided they are properly trained.

That’s where we in the Department of Supply Chain Management in the Harbert College of Business come in.