(© metamorworks – Canva.com)
When we think about the autonomous future, one of the first things that springs to mind is driverless cars. Like many, you might’ve balked at the idea of technology quite literally taking the wheel. Perhaps you still do. But self-driving vehicles are a good analogy for the technology that we should really be thinking about in the drive to intelligent automation and the autonomous enterprise.
I’m talking about a Digital Twin of an Organization (or DTO).
Instead of orchestrating the moving mechanical parts of a car, a DTO aims to autonomously orchestrate the moving parts of a business – its processes. By now, we’re well versed in the way digital twin technology replicates a physical object, with sensors enabling real-time data to flow between the real and virtual versions. Imagine having this capability for your entire organization and its processes.
You only have to look at current business challenges to see why a DTO is so valuable. In the age of artificial intelligence, where every business is looking to maximize the ROI of their technological investment, DTOs offer tangible value by supporting everyday operations and strategic decision making. When faced black swan events, such as the Covid pandemic, geopolitical conflicts, and a potential trade war, DTOs can help businesses rapidly model, evaluate and adapt. For example, a DTO could help companies identify alternative suppliers to mitigate the cost impact of tariffs.
But it’s this very type of disruption that means the DTOs your business may be contemplating aren’t entirely equipped to deal with today’s volatility. AI trained solely on historical data won’t be adequately equipped to handle new, unforeseen or extremely rare events. Instead, you need to upgrade your plans and build what I call a resilient digital twin.
How resilient digital twins work – and why they need process intelligence
A resilient digital twin can enable highly responsive decision-making and adaptability by providing a real-time, data-driven digital representation of an organization’s processes that can automatically react to shifting market conditions. But it can only do that with the right data and technology fueling it.
Process intelligence is the key to enabling the digital twin to make intelligent decisions, and we can see how this works with the Celonis Intelligence Platform. Using data from any source (ERP system, data warehouse, custom apps, etc.), Celonis uses process mining and augments it with business context to create a living digital twin of the business—the Process Intelligence Graph (PI Graph). It’s system-agnostic, without bias, and fully supports object-centric process mining (OCPM). The PI Graph shows how processes actually run, enables process optimization and feeds AI the input it needs to understand the business and how to improve it.
Because the structure of the object-centric event data powering the PI Graph is stable over time, the PI Graph is able to identify process optimization opportunities even when circumstances change dramatically. And when there isn’t enough training data, the Celonis Intelligence Platform gives back control to humans. We call this hybrid intelligence. Just like an autonomous car hands back control to the driver when the weather or road conditions require it.
Continuing with my autonomous driving example, those vehicles don’t just need an up-to-date digital model of the roads and the usual traffic patterns. They also need live information about congestion to determine the optimal route.
As a resilient digital twin, the PI Graph can integrate live data from trusted sources to provide additional context, drive even more informed decisions and orchestrate corrective action. In the supply chain, for example, a resilient digital twin can use external real-time data to detect a dramatic increase in suppliers’ pricing and automatically recalibrate the procurement process.
Without this data feed, all you have is a static process model – and one that doesn’t reflect the true picture of the organization at any one time. That’s of little use to you when you’re trying to keep up with daily disruption. With contextual data, the digital twin can recommend process optimizations to help businesses combat changing conditions in real time.
Digital doesn’t mean human-free
But there’s one other thing a digital twin needs to be truly resilient: ourselves.
No matter how sophisticated your training data and AI models are, disruption, emergencies and complex automation use cases require human intelligence. AI cannot know how best to respond to extreme or new situations that have never been experienced before, like geopolitical conflict or a global pandemic.
It’s like expecting the self-driving car to manage unprecedented weather or unusual terrain. Instead, the technology hands back control the moment weather conditions turn bad or the car leaves the highway.
We call this Out-of-Distribution (OOD) detection in the context of process mining. With a human in the loop, they can intervene when processes behave differently from historic data. Otherwise, we risk the digital twin making suboptimal decisions that are incompatible with how actual people and processes behave – and what’s best for the organization.
Implementing and maintaining resilient digital twins
Over time, I expect digital twins – like autonomous vehicles – will be able to handle abnormal situations as they become true DTOs. But this will be a gradual process. What’s important to do now is make sure your data management and process management are up to scratch.
The insights businesses gain from AI are only as good as the data and information it has available. The same is true of digital twins. Creating a DTO without first extracting data through basic forms of process mining is like trying to build a self-driving car without first implementing cruise control.
Once a feedback loop with existing systems is integrated into the digital twin, it can then start automatically triggering improvements based on process-mined insights. At Celonis, we see businesses implement automations across processes like Order-to-Cash, such as removing unnecessary payment blocks when they meet certain criteria. This means businesses aren’t just eradicating manual tasks, they’re also gaining the responsiveness they need to handle fluctuating conditions.
I started with what may have seemed a radical technological application in driverless cars. But they give us a blueprint for applying intelligent automation and AI to create resilient DTOs that truly improve long-term efficiency and agility in an increasingly uncertain business landscape. Always with the business in the driver’s seat.
