The word on the street is that artificial intelligence efforts have not been delivering the impressive results everyone has been hoping for. The payoff, however, is likely to be found in well-targeted placements in which AI is handling, in real-time, complicated tasks that would take humans days or weeks to unravel. Managing the flow of telecom traffic is one example, or keeping IT networks up and running is another. Still another, with enormous tangible business value, is the ability to keep supply chains flowing.
To that end, for example, Scale AI-- a Canadian consortium of companies, universities, and research centers -- announced $29 million in new investments in AI projects, much of which is aimed at building more intelligence into supply chains. One is an AI-based prediction platform for driving supply chain efficiencies for fleet managers, and another an innovative cloud-based platform for improving the drug distribution chain. A third initiative will enable the digitization of the mineral and metal value chain in order to secure supply chains and improve their performance."
Clearly, building intelligence into supply chain networks is a hot thing. One market estimate puts AI in the supply chain on a 46%-a-year growth track, growing from $503 million in 2017 to more than $10 billion by 2025. As Denis Forget, CEO of Distribution Pharmaplus, a participant in the consortium, puts it, AI-powered supply chains "will allow us to unlock productivity by improving inventory management, limiting the impact of shortages and reducing administrative management-a typical win-win solution, as we reduce our costs, while increasing our sales and offering better service."
What's at stake? Today's supply chain networks simply aren't nimble enough to handle today's and tomorrow's challenges, says Ram Krishnan, CMO of Aera Technology. For example, "manufacturers are now looking to produce smaller batches quickly and efficiently to address demand spikes across geographies and channels. Yet many are set up in inflexible, slow-moving manufacturing-at-scale models that aren't geared for agility and just-in-time batch production on demand. What was once a competitive advantage through economies of scale is now a barrier to nimble production."
What's conspicuously absent from today's supply chains "is the ability to adapt to ever-changing supply chain constraints including raw material availability, product and shipping times, and budgetary limitations," Krishnan continues. "Carefully planned lead times are upended when a disruption occurs at any link in the supply chain. If multiple disruptions occur simultaneously, the effects can cascade across the entire supply chain with potentially disastrous impact."
Look no further than the recent Covid-19 crisis, he illustrates. "Some sectors are grappling with far-reaching disruptions in raw material availability and shipping times -even as demand for certain products soar. They lack the strategic agility to course-correct as constraints change."
The issue that has been standing in the way is "monolithic transactional systems not suited to respond to rapid operational changes and enable informed decision-making at speed," he says. "Despite major advances in cloud architectures and database scalability, underlying batch-oriented supply chain systems have largely remained unchanged since the 1990s. Such archaic infrastructures make virtually impossible for supply chain practitioners to quickly get the right data to make the right decisions as disruptions occur and constraints change."
As a result, enterprises "are faced with a massive load of manual data work to collect information from disparate systems for analysis in spreadsheets or a data lake. Ultimately, best-guess decisions may be made days or weeks after an unforeseen circumstance arose."
AI can help provide the agility needed in today's supply chains, which are susceptible to all manners of disruptions, from pandemics to mundane hiccups in inventory software. "New challenges continue to emerge that make it difficult to achieve and sustain a truly agile supply chain," says Krishnan. He calls for greater "cognitive automation" in the supply chain, or the process of digitizing, automating and augmenting supply chain decision-making processes. This is supported through "a single cognitive data layer with deep data granularity by crawling internal and external applications thousands of times a day," he illustrates. This analytic layer can be built on existing IT infrastructure, and "monitors for changes, alerts practitioners to exceptions, and responds to any user query with a full set of options and ramifications."
Applying AI capabilities is the only way "to incorporate capabilities to handle crises systemically, provide end-to-end, real-time visibility across operations and augment human decision-making with machine-driven data collection and analysis," Krishnan argues. AI and machine learning algorithms need to be applied "to produce recommendations on optimal supply chain actions. Machines take over the heavy lifting of data collection and analysis at speed not humanly possible. Managers can focus on assessing AI recommendations and making swift, data-driven decisions to address disruptions, or simply make incremental improvements for overall efficiency."
With the help of AI, "decisions that took weeks of effort and meetings are being made faster with far greater precision," says Krishnan. "They're also improving orchestration across the full supply chain lifecycle, from planning and procurement to logistics and fulfillment. Harmonizing data from dozens or hundreds of applications opens new transparency into cause and effect across functions, and powerful what-if analysis. The depth of analytics goes beyond static rules-based reporting to generate a true picture of supply chain performance, weaknesses, and areas for improvement."
It's important to remember that "humans are not robots," Krishnan says. "We're not built to collect and process large volumes of information, certainly not at scale and speed. That's a job for machines."