AI Agents Automate Manufacturing Supply Chain Workflows & Create Opportunity
Industry 4.0 produced disappointing results. System-to-system AI agents are the final frontier for manufacturers to achieve real, large-scale cost reductions and savings.
Today's dynamic tariff landscape is introducing new challenges in manufacturing supply chains. When tariffs are imposed they don't just impose pricing adjustments once. Further cost and pricing impacts come when the initially tariff 'cost' is passed down to manufacturers who then pass this cost on to consumers.
Take aluminum tariffs, for example. When tariffs are imposed on aluminum, domestic manufacturers don’t just absorb the cost they pass it on to maintain their profit margins. This means everything that uses aluminum, from industrial robotics and car parts to hardware and equipment, and household appliances... becomes more expensive.
Price increases don't just stop at the initial tariff application. They build into a symphony crescendo as higher input costs influence further adjustments throughout the manufacturer's supply chain.
Higher prices influence the labor workforce to ask for higher wages so they can then keep up with the rising cost of living. This can push costs up even higher, which, in turn, forces businesses to raise prices again. Stagnation or deflation can be the result as demand plummets.
Manufacturing cost of goods sold (MCOGs) is about 75% of product cost. Mantras for better materials management and kanbans have been shouted from factory rooftops for decades with some success.
But materials management can only be taken so far with human activity.
BAAN vs inventory accuracy
I've lived through decades of manufacturers, vendors and suppliers trying to reduce costs by placing a fence around ERP and automating contract manufacturing environments with MES systems. In the mid 90s I worked in operations at Flex (Flextronics) and was part of the BAAN implementation team. The promise of ERP continuity was the early seed to Industry 4.0. At the time I thought we were leading innovation in manufacturing supply chains.
Mark Zetter
Regardless, post implementation third-party systems have compatibility issues and although they may accomplish some tasks they also have to be maintained. This opens the door for a third or even a fourth software vendor and often users still cannot update data or linkages. When this happens a manual work around with manual review processes are then employed as a manual update to the ERP system.
For BAAN, often parts were moved out of bins without proper paperwork transacted. Flex employees then checked production WIP and when they could not locate excess inventory transactions were reversed and then transacted, correctly.
Mark Zetter
The takeaway from that experience is if you need to use a third-party system, use it, but use it with its fullest intent and maintain it so system inputs and outputs are accurate.
IBM Watson vs IBM Watson AIOps 2.0
In 2011 IBM Watson won Jeopardy and artificial intelligence became a household term. But Watson failed to capture market share despite IBM throwing a ton of money at it. In 2018 I performed a IBM Watson deep dive and found no evidence IBM Watson was directly attributable to saving money for any IBM client.
In 2023, not one to give up, IBM launched IBM Watson AIOps 2.0 for which the verdict is still out and interest in AI continues to make headlines in all industries and markets.
Although most manufacturers already picked most of the low-hanging fruit for cost reductions and savings, namely, outsourcing for cheap labor and re-designed packaging, the largest market opportunities in manufacturing and supply chain cost reductions have remain untouched.
The biggest market opportunities for manufacturing and supply chain cost reductions and savings I see are in costly indirect labor workflows like S&OP, procurement, and compliance. Particularly with system-to-system AI agents.
I predict in 2025 we will see a mad rush by a lot of manufacturers to implement as much AI as fast as possible.
AI in manufacturing supply chains
Today, most AI works 'with' people to speed up workflows. This comprises 90% of company activities according to market intelligence firm CB Insights. The technology is "serving as a stepping stone to more autonomous solutions."
I've had dozens of conversations with vendors of AI agents and manufacturing automation plus 100s of conversations with different manufacturers over the past 18 months. Every firm want to grow income or save money.
There is tremendous market potential - into the next decade - for vendors to capture market share in nearly every industry and market vertical and, particularly, in EMS manufacturing industry with its many unwritten codes of human conduct woven into relationships between in-house functional groups supporting production and 100s of 1000s of vendors and suppliers to OEMs and EMS and ODM providers with extended contract manufacturing supply chains.
System-to-people agents vs system-to-systems agents
There are primarily two types of AI agents: system-to-people (platform interface) agents and system-to-system agents. As CB Insights proclaimed, today's agents are primarily system-to-people. This is because true system-to-system agents have not yet been fully perfected despite what many vendors claim.
There are a ton of AI agent companies jockeying for the manufacturing supply chain, including product design, import and export compliance and, so on. Some have good ideas but they cannot execute because of uninformed buyers and many are not narrowing their focus enough while others are not solving the right problems.
Some questions for manufacturers to think about when looking for AI solutions; how do I identify problems suitable for multi-agent systems? How do I know which vendors can deliver agent solutions we need? Where should agent(s) be implemented in our supply chain? Which functional groups can benefit the most? Which activities and how can a multi-agent system benefit our supply chain? How do we assemble our agentics project team and monitor vendor progress?
AI agents bring opportunity
CB Insights has stated, "AI agents could manage entire industrial processes, shifting human roles from operational tasks to strategic oversight." I could not agree more. Identifying areas for deploying AI agents and communicating knowledge about the benefits of AI agents is becoming extremely important.
Mark Zetter
The impact to SaaS and software industry will be remarkable. According to Aaron Levie, CEO @box,"In a world of AI Agents, clearly this is going to be very different. Agentic workflows have no upper limit on how much they can be deployed by an enterprise. And all of a sudden the software categories that were once constrained by seat volume, have no such limits anymore. We're already seeing examples of AI Agents in coding, research, legal work, and other advanced categories that are being billed at multiples of their prior seat-based software equivalent price."
Also, from the same, longer post on X, Levie writes, "This provides a completely new growth vector for software companies in AI, and has major implications to software business models."
Clearly, there are market opportunities and opportunities for jobs.
The takeaway for manufacturing and supply chain professionals reading this is not that AI can replace jobs. A deeper understanding is knowing working with AI can help in your career development and increase your value while creating opportunities.
How do you know if AI agents created for manufacturing supply chains are good or not? First, you have to understand the unwritten codes of humans working in electronics manufacturing supply chains. You then have to test the AI agents. Talk to the agents and evaluates agent output beyond conventional benchmarks. Read more about my manufacturing approach to AI here.
_________________________________________________________
About
What matters when formulating contract electronic strategy? How do you identify supplier profit centers and what are you doing to protect against margin erosion for your outsourcing programs? Why do provider capabilities often not match capabilities they claim? How are you benchmarking your supply chain against competitors?
I’ve spent 25+ years in contract electronics industry setting up contract electronic divisions and running operations, protecting EMS program profits, manufacturing capacity M&A and more. I run a technology solutions firm. A lot of times this means asking the right questions.
_________________________________________________________