Artificial intelligence may compensate for the opaqueness of supply chains

How soon until artificial intelligence (AI) sharpens up our supply chains? It won't be long. A recent survey, conducted by Forbes Insights, SAS, Intel and Accenture, finds 36% of companies are adding AI capabilities to their logistics and supply chain operations with another 28% planning to do so. 

The challenge with supply chains has been their opaqueness. After all, they involve a lot of moving parts, each run by different entities across different continents using different systems. "Supply chains remain far from the ideal of real-time visibility across the supply chain operations, hence, data-driven, accurate decisions," states Fred Laluyaux, CEO of Aera Technology. This is the case despite years of implementations of "an array of transactional and analytic systems for supply chain operations -- no matter it's on-Enter premises or cloud."

That's because, as with everything else touching enterprises, supply chains "have grown much more complex in today's faster digital environments. Soaring data volumes and diversity are overwhelming. More partners, products, geographies complicate, global competition and rising customer expectations complicate the dilemma." 

Add one more element to the pressures of supply chains -- the Amazon effect. Companies really need to tighten up their processes to the point where moving products from factories to warehouses to distributors to customers' doorsteps is quick, visible and painless. 

Still, "too many supply chains run on guesswork decisions based on information that's often outdated and contradictory," Laluyaux warns. "Companies risk delays, needless costs, and revenue loss by continuing to rely on status quo processes and decade-old software."

Enter AI. "AI relies on machine learning algorithms to learn and refine in real time as it crawls internal and external data sets," he continues. "That could include inventory data, supplier performance, demand fluctuations, and even weather or road conditions. AI combines this disparate knowledge to make recommendations or decisions on optimal actions. Think of it almost as a self-driving business based on cognitive automation—the ability to learn, think, and take actions."

Laluyaux points to the available-to-promise function, which responds to customer order inquiries based on resource availability. In traditional software, available-to-promise "is fundamentally a rules-based calculation based on theoretical lead times and allocation rules that are incredibly variable and volatile," he says. "Using those data points in calculations can result in wrong dates." 

In contrast, he explains, "AI can automatically generate a supply-chain map, showing everything about an order, including allocated quantity and expected delivery date. It delivers highly accurate recommendations and predictions based on machine learning and data science, not simple rules-based calculations."

Beyond available-to-promise, AI's cognitive automation capabilities can be applied to all supply chain processes, Laluyaux continues, "from demand and supply forecasting to inventory optimization, manufacturing performance, procurement automation, and supplier reliability assessments." He calls this a "self-driving supply chain." He sees the journey to AI-enhanced supply chain going through five levels of discovery and development:

1. Understanding. "Leverage AI to fully understand the true operative states of your supply chain."
   
2. Recommendations. "Utilize the AI system for recommendations on key risks and opportunities."
   
3. Predictions. "Gain insights with predictions and probabilities based on AI's continuously evolving machine learning."
   
4. Augmented decisions. "AI suggests optimal decisions that require human review and approval."
   
5. Autonomous decisions. "AI makes decisions autonomously without human intervention."
   

AI is a journey, and it's greatest value comes from handling complexity. In the business world, nothing gets more complex than supply chains, making this a ripe zone for AI. 

(Disclosure: I was part of the team that produced the Forbes Insights-SAS-Intel-Accenture survey mentioned at the top of this post.)