The Insanity Behind AI-Enabling your Supply Chain

It’s been said that repeating the same thing over and over and getting the same result is the definition of insanity, read on…

Most product manufacturing, distribution or retail companies consider their supply chain management process to be their most important value creation process.  And they make it a priority to continue to invest in the technology and human resources that operate the supply chain. 

These investments don’t always work out.  One bright spot has been efforts to continually improve supply chain performance using iterative six-sigma techniques to define and communicate best practices.  However, even these efforts tend to produce diminishing returns if they focus on the same KPI or the same workflow for too long.

Eventually, most companies decide that they need to invest in new software and technologies to continue improving performance.  In the past, many companies invested millions or billions implementing ERP solutions that usually did not live up to expectations.  No need for us to review the history of supply chain reengineering and ERP failures that cost so many executives their careers.

Today, the new paradigm for improving supply chains is to AI-enable the supply chain.  The goal being to use machine learning to analyze complex datasets and provide near-real-time insights or optimal decision recommendations to supply chain knowledge workers. Successful early adopters have shown that this recipe works.  McKinsey reports (see article) that:

  • Logistics costs can be reduced by 15%
  • Inventory levels can be reduced by 35%
  • Service levels can be improved by 65%

So, what are companies doing to AI-enable their supply chains? Ironically, they are looking for solutions from the same leading ERP vendors and other large supply chain solution providers that failed to deliver the results they needed and expected in the last round. The same vendors selling multimillion dollar behemoth solutions that present the exact same challenges seen before.

In the same McKinsey article noted above, a very alarming history of challenges and disappointments is reviewed that makes the implementation of these new AI/ML applications sound a lot like the failed promises of the last ERP reengineering cycle.

From thirty thousand feet, large scale, expensive, ‘one size fits all’ supply-chain planning solutions fail to meet expectations most of the time because:

  • More than 60% of implementations are over budget or late
  • 35% failed significantly to deliver the impact they promised

Further, the main implementation challenges identified in the report sound a lot like the ERP challenges of the recent past:

  • Change management difficulties (82%)
  • Poor process design (70%)
  • Missing capabilities (57%)

More specifically, McKinsey identifies problems that must be avoided to ensure success.  Again, these problems may sound like they were copied from an ERP implementation that you supported in the past.

  • Value was not clearly identified: less than 1/3 of companies perform a value diagnostic
  • Design phase was overlooked: most companies choose partners and solutions before completing the end-to-end design
  • Insufficient focus on execution
  • Only 25% of executives believed that their objectives and their system integrators objectives were well aligned
  • Inadequate capability building (i.e., knowledge workers don’t understand how to use the new system): only 13% of executives felt their companies were adequately prepared to address the skill gaps created

Are you destined to repeat the same mistakes?

One unstated assumption of the McKinsey article is that the new wave of AI/ML ERP that connects a company with all its suppliers and customers is clearly the way to go.

What if this assumption is wrong altogether? Considering the risks and challenges experienced in the past with large, expensive, ‘one size fits all’ solutions, isn’t it worth considering other options?

One option is to consider a more subtle, gradual approach to the use of AI solutions in your supply chain processes. A thoughtful, problem-solving approach would allow consideration of your company’s specific business rules and knowledge worker daily workflows to identify when and where data and AI/ML is able to improve performance.

The McKinsey report mentions a lack of a design phase as a high-risk problem for the conversion to an AI-enabled supply chain. In other words, many companies do not first understand what they need versus the shinney objects they are told they need in many vendor offerings. In short, they don’t take the time to understand where, in their specific supply chain planning process, AI/ML will benefit them.

The McKinsey report also mentions that executives and IT are often out of step in terms of project objectives and that knowledge workers are not considered in the design and left with systems they can’t use or do not want to adopt. These problems are sure to be repeated if supply chain executives continue seeking out the same failed  approaches to addressing an ever-changing global supply chain.

A new paradigm should be considered which uses a more sophisticated, targeted ‘design approach’ to AI-enabling your supply chain operations. Without a design plan fit to your business, you will be repeating the same ERP challenges of the past.


That is what A2Go believes, and it is this design philosophy that drives our ability to deliver targeted AI applications to any supply chain process.  Our AI rePlan solution consists of a proven process for the quick evaluation of each customer’s challenges and the AI building blocks required to build customer specific AI supply chain portfolios.

How are we different?

  • No big-bang approach; we focus on incremental AI-enablement of the supply chain
  • Knowledge workers help us identify and prioritize pain points in the supply chain where the AI/ML techniques could help most
  • No requirement to change process steps or supporting workflow automation
  • Step-by-step implementation of interoperable AI applications with the timing and sequence agreed with knowledge workers in advance

A2Go provides a practical alternative to the mad rush to AI-ERP oblivion.  Please take the time to learn more about our solution and consider us as an alternative to the madness.

Related Posts

About Us
We set our goal to make it possible for businesses to use data to optimize recurring, mission-critical decisions in real time and at the point of use – by sales teams, pricing teams and supply chain management teams.

Let’s Socialize

Popular Post