How Intelligent Agents can Solve Supply Chain Problems

It used to be that each part came from a single manufacturer and, often, internal BOMs (Bills of Material) and drawings to make products used that manufacturer’s part number. Also, such parts could be reliably ordered from distributors on just a few-days-notice.

Then, suddenly, parts that used to be readily and reliably available were/are no longer available as factories in China have closed or moved to elsewhere in Asia or Mexico. In the short-term, some of this mayhem was driven by the pandemic and more recently by tariff wars. But in the longer term this has been much more driven by supply chain managers realizing that, at $5/hour, China’s labor rates were no longer competitive with those of other countries, at under $2/hour. As a result, the supply chain shifted to different countries and became much more complex.

Now sourcing parts has become a scramble.

Ideally, if the part from a primary supplier is not available, when needed, then interchangeable or alternate parts may be available. But typically, the replacement parts will have different part numbers. So, the BOMs and drawings for making assemblies or other products, which specified the primary supplier part numbers are now incorrect and need to be changed.

This situation is made worse, if only substitute parts are available, which only work in specific situations, or even worse require changes to the assembly or product in which they are used. We have even seen companies stripping parts from old assemblies and refurbishing them so they can be reused, when no other substitutes are available with less than a 6 month lead time.

Managing all of the resultant engineering change orders (ECOs) can quickly turn into a paperwork nightmare if not delegated to intelligent agents, for example, to automatically change kitting and work orders, and possibly drawings, to reflect available part numbers rather than simply referring to the preferred supplier part number.

Then there is the issue of parts delivery. Previously, delivery of needed parts for production could be scheduled a few days ahead of when needed. Now parts may be randomly delayed without warning. This requires that production work order steps be dynamically scheduled based on available parts as well as customer delivery priorities. This, again is a complex task best left to a scheduling algorithm in an intelligent agent.

With such volatility, it is important to communicate automatically with supplier systems to periodically get updated delivery status information. This is to provide the decision support information needed by materials managers, when ordering needed parts based on the materials planning being performed dynamically by other intelligent agents, in response to the incoming flow of customer orders.

It is also critical, in such circumstances to keep customers and downstream distribution warehouses updated as to the status of their orders. This is to avoid customers endlessly calling customer service and expediters having to roam the floor to find the status of customer orders. Again, this is a task best left to intelligent agents which can automatically detect when changes have occurred or are planned to occur and send text or Email alerts to customers and update systems used by downstream supply chain partners.

Back in the “Good Old Days” of make-to-stock manufacturing, it was possible to manage a company’s supply chain with an ERP system, or even an accounting system, supplemented by Excel spread sheets and paper forms. But now the level of complexity has increased by an order of magnitude or more. Also, the time available to make decisions has shrunk to hours or even minutes. As such, it is now becoming impossible for even a large team of people to manage industrial supply chains without the aid of intelligent agents to do much of the needed real-time “Intelligent Grunt Work”.