Accurate forecasts are possible in the current market with Enterprise AI
Earlier today, Tyson Foods, one of the largest food processors and marketers in the world, issued stark guidance to the market. The headline read “Current market makes accurate forecast impossible.” As a result, for the first time in decades, Tyson declined to provide sales guidance to the analyst community.
From the article: “The reasons Tyson cited include ASF (African swine fever), shifting trade conditions, and a fire at a key facility.” The executives said the factors blocking sales forecast are largely external and hard to predict.
To understand Tyson’s predicament, we have to understand how complex the modern supply chain has become and the role that signal-to-noise ratio plays in the planning process. We’ve illustrated this in the animation below.
The modern supply chain is a case study in the benefits of specialization
In response to largely inefficient processes and production difficulties, the 1980s ushered in supply chain management (SCM) as we know it today. Every aspect of the supply chain was segmented into a specific activity (e.g. planning, selling, production). People would be trained to do only one thing and do it exceedingly well.
The 1990s accelerated this specialization with the advent of enterprise resource planning (ERP) tools that mimicked the physical structure of the supply chain. These are the tools that many companies still use to date. However, the consumer of today is nothing like the consumer of decades past. Since all supply chain is driven by consumer behavior, companies like Tyson find themselves in the difficult situation of understanding and simulating external realities with a limited toolkit that was only meant to operate in a static and noise-free world. We all know that’s not how the world works!
Traditional software (such as ERP systems) serve as great systems of record. They do a fantastic job of instrumenting the flow of data from various points in the supply chain. Unfortunately, that’s where the usefulness ends. The ERP systems fall short in providing intelligence at the point of decision-making.
There are several reasons that explain these limitations. The first is that most traditional software is constructed as a series of rules-based engines. These static rules (or heuristics) don’t learn or get updated until a programmer changes the code. The second is that traditional software only understands what happens within the four walls of the company. It has little to no knowledge of market realities or shifting conditions, as Tyson described. The third difference is more nuanced – traditional software does not understand probabilities. It operates in black and white.
With these limitations, there are many more companies that feel pain in their supply chain due to a lack of intelligence at the right decision points. These are precisely the types of problems that are tailor made for AI/ML algorithms. When balancing between thousands of variables using limited tools, it can feel like you are playing a game of 3-D chess with a blindfold on. You need have the machine on your side! That’s where we come in.
We started Noodle.ai with a simple goal of reducing waste in the supply chain by addressing the limitations of traditional software. Far too often we hear of missed fill rate opportunities causing lost sales or excess inventory sucking up the working capital of a company because the planning process was biased. why was it biased? Because the tools could not help users think in a probabilistic way, or because an artificial black and white choice was imposed on the planner. Absent intelligence, most planners will resort to the law of averages and leave money on the table.
Jeff Bezos famously said the secret to Amazon’s success is in constantly using data to do better than the average. We agree.
There’s a better way to do Supply Chain. Just ask our customers.
As for the headline of the Tyson article? We agree with it – with a slight twist: “Current market makes accurate forecast impossible IF you rely only on traditional software.”
Header image from Supply Chain Dive
- Introduction to MLflow for MLOps Part 3: Database Tracking, Minio Artifact Storage, and Registry
- Believe the Hype: Gartner, Noodle.ai, and Decision Intelligence
- Introduction to MLflow for MLOps Part 2: Docker Environment
- Atlas, Noodle.ai’s Machine Learning (ML) Framework Part 3: Using Recipes to Build a ML Pipeline
- Supply Chain Leaders Struggle with Unpredictability. Noodle.ai has the Cure.