It may seem that today’s increasingly unpredictable supply chains have finally bested professional planners, but the reality is that legacy planning software hasn’t caught up to our ever connected and interdependent world. Traditional tools rely on rules-based prescriptive solutions: We make some educated guesses, draw a linear regression on a spreadsheet, and cross our fingers. Deviations from those predictions are “exceptions,” and we manage them with buffers or expedites — locking up working capital, misdirecting production cycles, and frustrating consumers with empty shelves.

There's a better way: restoring perfect flow to supply chains.

The opposite of perfect flow is a phenomenon we call “operations entropy.” It’s the disruption of well-laid plans by forces that are generally thought to be unpredictable. With advances in computers, data storage and machine learning, operations entropy is finally being defeated. Even if your supply chain is starting to leverage AI advancements in prediction, it's easy to go astray. Keep these points in mind to move your organization forward:

Focus on value at risk. Alerts about lack of supply and excess inventory are just distractions from the most critical number in planning: value at risk. All too often planners spend their weeks weeding through alerts without the context or visibility necessary to make informed decisions. Put some compute behind those alerts and quantify the emerging risks to give your entire team visibility into exactly which problems are putting the most dollars on the line. It may still take some time and research to mitigate a thorny issue, but everyone will rest assured knowing they’re working on exactly the right problem.

Don’t settle for anything in a black box. Trust is critical in a high-stakes and maddeningly complicated supply chain breakdown. No AI will have the human intelligence that a planner has, but the AI will have far more rigorously interpreted context than the planner. AI platforms need to expose their reasoning to planners to verify the assumptions, and to give planners the confidence to act on those insights. Making large impactful decisions requires courage and assurance, not just big data and clever algorithms.

Eliminate waste. Waste is the outcome of inaccurate predictions: Inventory buffers mask imprecise demand forecasts, and empty trucks bely logistics oversights. Even factory downtime and defective products can be prevented with better predictive science. Many of these inefficiencies are accepted “business as usual,” but with new predictive tools, it’s time to question every source of waste — like inventory, overproduction, transportation, or even wasted talent.

Turn down the noise. The accumulation of legacy systems of record often adds unnecessary complexity to a planner’s digital workspace. In the course of mitigating a disruption it’s common to have multiple log-ins, redundant alerts, and at worst permissions or visibility issues. AI excels at data manipulation, tagging, categorizing, and correcting many of the data streams that legacy systems produce. AI solutions should make all the relevant data readily available, so planners can focus on mitigating a disruption, not sorting through tangled systems for better visibility on the issue.

Embrace collaboration. The future is human and machine intelligence working together. AI and machine learning applied to supply chain planning is not some future state on an analyst’s hype cycle. The value is proven in production — with the goal of pushing past hyped expectations and seeing real business value.

Supply chain planners didn’t need a pandemic to point out what they already knew: The speed of decision-making within the execution window is critical for supply chains. Now is the time to support supply chain planners and leaders with products built with AI at their core.

Mike Hulbert is vice president of consumer business at Noodle.ai, an AI products vendor for the supply chain and manufacturers.