A revolution in supply-chain technology is "on the horizon," says Paul Blake, director of technology marketing with GEP. He explains the nature of that change, and what's driving it.

SCB: What is driving the need for new supply-chain technology?

Blake: It’s about changing circumstances for large, complex businesses. What they're seeing across the supply chain is increasing volatility and uncertainty. Factors that have always affected supply chains are being more keenly felt. Political, social, environmental and market changes are having much more of an immediate effect, with deeper impact. Our customers are looking for new ways to get ahead of the game, gain competitive edge, and assert better control over their supply chains, so that they can react more efficiently in these changing times.

SCB: What types of supply-chain technology are available right now to address these needs?

Blake: Fundamentally, it's all about data. The existing technologies that our customers have to rely on are largely around legacy systems, enterprise resource planning and the like. There’s a lot of disconnection between systems, functions and business units. Supply-chain operations are siloed. There’s a great opportunity for driving value, if you can apply all of this data and intelligence to a control tower, which gives you command and control over the whole supply chain. What we're seeing is a new paradigm, using artificial intelligence as a means of accessing huge volumes of data in real time.

The AI part isn’t particularly new. Some of the algorithms and machine-learning code are decades old. What’s changed in recent times is access to vast volumes of data, in a “data lake” model, along with the computing power to run those algorithms many of millions of times in various iterations. You’re able to create a digital twin of operations, and achieve new levels of visibility and command and control.

SCB: How mature are these AI systems, in being able to output the necessary decisions that will actually work in the organization? Do they still have a ways to go?

Blake: Absolutely. It's important to stress that core AI capabilities are fairly mature. What’s new and takes time to come into maturity is how you combine vast volumes of data with those AI algorithms, and package them in such a way that an organization can effectively access all of the intelligence that’s produced. That’s the piece that doesn't exist today, and it’s what we're seeing as the next step in supply-chain technology evolution.

SCB: What about ensuring “clean” data? Does the AI engine make that determination before it even starts to manipulate the data?

Blake: Yes, to a certain extent. There’s a stratification of the model, with the data at the bottom, and the data lake being the fundament of what you're working on. Then there's a layer of intelligence that sits over that and has to interpret the data. Some of that intelligence is artificial, but a good percentage has to be human as well. There has to be an understanding of whether the output from the AI is meaningful.

SCB: We have to determine whether the output from the AI system is predictive or prescriptive. I would think the latter is a lot more rare.

Blake: It certainly is today. We see it as an evolutionary path. The first step is to get some kind of predictive model that you know can be trusted. Is that prediction of demand for your product correct? As the algorithms become more enriched with internal and external data, those predictions become more accurate. Only then can you take the next step to prescription.

SCB: Is the system capable of explaining its conclusions, or is it a black box that just spits out results, and you have to trust them over time?

Blake: It’s very hard. A lot depends on how you create the visualization layer, the app layer on top of the intelligence. As I said, there's a stratification of data, intelligence, and the app. A lot of the skill in software development for the future will be in creating the user-experience layer.

SCB: You’ve said there's a revolution in supply chain technology on the horizon — meaning it’s not here yet?

Blake: For some organizations that we’re working with right now, we're seeing some immediate opportunities coming to fruition. For others with a heavy, embedded legacy of existing systems, or a lot of political and organizational inertia, it’s going to take time before they can take advantage of this technology. But certainly right now, we say to our customers, "If you give us a lot of data and a little time, we can show you some incredible results."