3 Solutions for Common Supply Chain AI Risks
What comes to your mind when you think about supply chain AI?
If you imagine robotic warehouses, autonomous vehicles, smart systems, and Industry 4.0, this post is for you. If your vision included red flags, hype, and unrealistic expectations, this post is for you too.
There are three red flags you need to know about, plus solutions for keeping your supply chain AI projects out of the red.
Red Flag 1: Urgency
This flag is on just about everyone’s list, along with AI. After nearly two years of battling a supply chain wildfire, hope in AI’s salvific power is reassuring weary souls longing for peace of mind. A recent report from Gartner found Artificial intelligence (AI) / machine learning topping technologies, with 48% of CIOs and technology executives having deployed or planning to deploy these technologies in the next 12 months.
The high urgency, expectations, and buzz surrounding AI beg for healthy skepticism. Like a preflight check, urgency isn’t a waiver for what’s important - especially if there’s a risk of negatively affecting multiple enterprises. The cost of acting in haste can far outweigh perceived benefits, as the next two red flags show.
Solution: Follow an implementation plan and focus on what’s important. Explore other solutions for resolving urgent matters that don’t compromise critical goals in the plan. Notify others, in writing, of specific risks that may result from changing the plan.
Red Flag 2: Unrealistic Expectations
Unrealistic expectations are the beginning of many troubles in life, including technology project failures. The current environment contributes much to this problem with supply chain AI.
85% of AI and machine learning projects end in failure according to a study by Gartner Research. AI requires proper planning, implementation, and training, just like any other technology. Like most newer technology, AI comes with a steep learning curve.
Part of the learning curve involves setting realistic expectations about AI. Many buyers believe that a business merely provides systems access and funding, and the AI vendor does the rest. Fuzzy talk about prescriptive analytics solving logistics problems and identifying process improvements is common. The best way to curb runaway expectations that can lead to project failure is to educate others on AI’s limits.
In short, AI acts as an agent for us, which is the great potential and problem with AI. It mirrors, amplifies, and projects whatever a company wants, for better or worse. AI can enable a well-designed process to achieve more or take a less effective business further down the wrong path with greater risk. More about this anomaly in the next red flag.
Solution: Clearly define expectations from the start. Confirm a vendor’s proposed deliverables meet your requirements for functionality, performance, and timelines. For SaaS cloud solutions, find a vendor who demonstrates proof of concept using a sample of your company’s actual data. Educate others on what to expect from AI.
Red Flag 3: No Supply Chain Model
This will be the downfall of countless AI projects. Most companies don’t realize this because a model wasn’t as important before AI and digitalization became star attractions. It is my personal opinion awareness will grow over the next several years, especially as supply chains seriously grapple with the Supply Chain 4.0 riddle.
An effective supply chain model is the optimal starting point for effective AI. Likewise, continuous improvement of supply chain processes is the best strategy for continuously improving AI. AI is not part of the model - it’s a technology for boosting the performance of processes defined in your model.
A model is a tool that makes it easier to align technologies like AI with corporate objectives: strategy, goals, values, priorities, policies, methods, and culture. It presents enterprise workflow, data, expectations, and supply chain partners in alignment with these objectives.
Decision-making data and logic that govern an operation become the rules for AI. A good model aligns everything, including these rules. The end result is a frictionless, fully integrated supply chain with flawless transparency. Wowza.
Here’s the reality of AI. AI processes any data and rules you feed it regardless of the system(s) behind it. It doesn’t care whether the host enterprise is the technical perfection, or a horror show of legacy systems. AI will consume underlying systemic dysfunction like sugar-coated poison pills and silently undermine the health of a business. Here’s how AI can escape detection.
Enterprise KPIs and metrics will improve as AI optimizes a process, but systemic flaws remain undetected. These problems introduce friction to enterprise processes, and they limit optimization potential. A business with a less effective supply chain model cannot match competitors’ efficiency gains with better models’ AI. Over time, relative losses compound, eventually revealing themselves in the marketplace and company financials.
Unlike any technology employed in the past, here’s why AI and a good supply chain model are crucial for long term business viability:
- AI’s optimization increases the gap further and faster between competing supply chain models.
- Optimizations compound - enabling further process improvements with each degree of optimization.
-
The network effect will kick in once AI is employed to optimize processes across multiple enterprises, increasing gains to unprecedented levels.
Solution: Create a holistic process model for your supply chain, inclusive of supply chain partners. Your data model can also become the digital twin for your supply chain. I suggest a shipment-centric model for achieving end-to-end control across multiple enterprises
Conclusion
Do red flags come to your mind when you think about supply chain AI? If they do, then robotic warehouses, autonomous vehicles, and Industry 4.0 are far more likely to be part of your future.
No matter what you envision with AI, I trust this post was for you too.
