Without a doubt, supply chain decision-making is tricky. First, supply chains increasingly are becoming more complex. Moreover, unplanned disruptions continue to escalate, hammering fragile logistics networks. Meanwhile, our enterprise systems that should enable supply chain executives to make rapid, informed decisions, don’t. In fact, the root cause of this system disconnect lies in the fragmented nature of supply chain data, which is often scattered across multiple systems and stakeholders. To overcome this, businesses need a more holistic approach to supply chain analytics. Specifically, they need a capability that unifies disparate data sources and provides actionable insights, on-demand. The silver lining is that emerging technologies like Multi-Hop Reasoning are now available to enable rapid, informed decisions.

In this article, I’ll share my insights on how Multi-Hop Reasoning technology is a real breakthrough for supply chain decision-makers. First, I’ll identify the top three shortcomings of current IT systems that impede executive-level decision-making. Then, I’ll provide a primer on how Multi-Hop Reasoning works, empowering better supply chain decisions in this age of AI. Also, I’ll explain how Multi-Hop Reasoning integrates with emerging technologies like AI and Knowledge Graphs to make it a powerful IT capability for decision-makers . Most importantly, these decision system capabilities are available to supply chains today to enable rapid, informed decisions.

1. Supply Chain IT Shortcomings that Drive Poor Executive-Level Decision-Making.

As supply chain executives navigate the complexities of modern business, they’re often constrained by IT systems that fail to provide timely, accurate, and actionable insights. Consequently, ill-informed, haphazzard decision-making is the norm for most supply chains. To explain, below are the top three shortcomings of supply chain IT today that are causing these information gaps. These shortcomings include the following: rigid enterprise software silos, lack of quality data for timely decision-making, and bloated, feature-rich systems that do not actually support executive decision-making, resulting in unintended consequences.

“Consequently, ill-informed, haphazzard decision-making is the norm for most supply chains.”

a. Enterprise Systems Are Rigid, Monolithic, and Hoard Data.

Agile Supply Chain Decision-Making: First You Need to Know The Truth About Enterprise Software

First, let’s look at enterprise systems such as WMS, TMS and ERP to name a few. Over the years, software developers have designed these systems for their intended purpose to be robust, transaction-oriented, and reliable. However when it comes to executive decision-making, these systems lack flexibility and agility to enable decisive action in our ever-changing supply chain landscape. Undeniably, these enterprise systems are rigid and monolithic. Moreover, these systems tend to hoard data, limiting visibility and collaboration across the supply chain. For more information on this topic, see my article, Agile Supply Chain Decision-Making: First You Need to Know The Truth About Enterprise Software.

b. Supply Chain Data is Disjointed, Duplicated, Ambiguous, Inaccurate, Out-of-Date and Incomplete.

Data-Centric Advice to Reduce IT Complexity and Make Tech Remarkably More Useful

Without a doubt, supply chain data for decision-making is often inaccurate, incomplete, and outdated. This is particularly true at the executive level where complex information requirements span multiple functional areas like finance, operations, sales, and planning. The root cause of bad data in our supply chain is a consequence of our application-centric mindset. This is where we prioritize, even glorify, software applications, making data just a by-product. Consequently, we have poor quality data, resulting in ill-informed decision-makers. Moreover, even AI will not compensate for this, as it depends on accurate data to produce reliable outputs. Undeniably, this is a case of “Garbage-In, Garbage-Out“.

For more information on this topic, see my article, Data-Centric Advice to Reduce IT Complexity and Make Tech Remarkably More Useful.

“The root cause of bad data in our supply chain is a consequence of our application-centric mindset. This is where we prioritize, even glorify, software applications, making data just a by-product.”

c. Unintended Consequences Due to IT’s Lack of Focus on Supporting Executive-Level Decision-Making.

Unintended Consequences: Everything You Need To Know With Examples, Causes, And Remedies

Lastly, complex supply chain systems have created a challenging landscape that impedes rapid, informed decision-making. Moreover, this digital complexity results in our supply chains being overwhelmed by unintended consequences. In particular, our IT systems are often bloated and focused solely on transaction processing or planning, failing to support executive-level decision-making. Further, descriptive Business Intelligence (BI) dashboards are at best a band aid to help with decision-making. As a result, we have significant information gaps for both department-level managers and senior executives. As mentioned previously, AI-powered technology will not overcome this culture of unintended consequences, if we do not have IT systems that directly focus on executive-level decision-making.

For more on the nature of unintended consequences, see my article, Unintended Consequences: Everything You Need To Know With Examples, Causes, And Remedies.

2. A Primer on How Multi-Hop Reasoning Works to Enable Better Supply Chain Decisions in the Age of AI.

As I have shown, supply chains face unique challenges in leveraging technology that directly supports decision-making. However, there are emerging capabilities in AI and Knowledge Graph tech that offer promising solutions. One of these solutions is Multi-Hop Reasoning that can help our industry overcome its biggest hurdle: fragmented data. This is because Multi-Hop Reasoning tech can enable on-demand analytics across multiple tiers and stakeholders’ systems. As a result, decision-makers can better identify potential bottlenecks, anticipate disruptions, and make timely, informed decisions. Below, I’ll explain what Multi-Hop Reasoning is and how it can enable smarter, more reliable decision systems for complex problem-solving.

a. What Is Multi-Hop Reasoning and Why Do Supply Chains Need It for Informed Decision-Making?

First, let’s start with a simple definition for Multi-Hop Reasoning:

“… can be defined as a process by which conclusions or answers are derived by sequentially combining information from multiple pieces of evidence.”

Simon Welz

For a graphical example of how Multi-Hop Reasoning works, see image below that both depicts specific attributes of the song, Cloudburst (in BLUE) and asks a complex question about it. More specifically, this illustration depicts a Knowledge Graph with entities as the nodes and the relations as the edges, mapping out a 2-hop question answering process.

From a supply chain perspective, this multi-hop analytical reasoning is crucial in this age of AI. This is because AI needs a way, on-demand, to leverage the increasing number of disparate supply chain systems containing our critical data. Moreover, this method of analytics coupled with the right tech will ensure that supply chain decision-makers rapidly get the insights and visibility they need to make informed decisions. Indeed for supply chains, Multi-Hop Reasoning goes far beyond data integration between our many supply chain systems, owned by numerous suppliers, manufacturers, logistics providers, financial institutions, and customers to name a few. What supply chain decision-makers need is advanced tech that provides on-demand, synthesized information at their points of decision.

“What supply chain decision-makers need is advanced tech that provides on-demand, synthesized information at their points of decision.”

b. Leveraging AI Using Multi-Hop Reasoning, Multi-Hop Question-Answer, and Chain-Of-Thought Prompting.

Surprisingly, it is just recently that Multi-Hop Reasoning has acquired the capability to be a powerful tool for operational decision-making. This is mainly thanks to emerging technologies such as AI and Knowledge Graphs. Previously, reasoning for complex problems was too slow and laborious for rapid decision-making. To get a better understanding of this new analytical reasoning capability, let’s examine how Multi-Hop Reasoning supercharges AI and other emerging technologies. Indeed, this new capability enables swift and informed decisions in complex environments like supply chains. To start, I’ll first define key multi-hop reasoning methods and techniques. See below.

Definitions of Key AI Multi-Hop Reasoning Techniques

• Multi-hop Reasoning. “The ability of an AI system to access and process information from multiple sources to answer complex questions or solve problems that require connecting the dots across different pieces of information” (SymphonyAI).
• Multi-hop question answering (MHQA). This technique is a sub-capability of multi-hop reasoning and a sub-field of Natural Language Processing (NLP). This method can be used with Large Language Models (LLM) as well as with Knowledge Graphs “ … to perform complex reasoning and navigate through a knowledge base or a collection of interconnected documents, to find the correct answer. In essence, MHQA requires multiple “hops” of information retrieval and inference to connect the dots and arrive at a comprehensive response” (Wisecube).
• Chain-of-Thought (CoT) Prompting. This AI prompt engineering technique enables multi-hop reasoning by enhancing “… the output of large language models (LLMs), particularly for complex tasks involving multistep reasoning. It facilitates problem-solving by guiding the model through a step-by-step reasoning process by using a coherent series of logical steps” (IBM).
• Retrieval-Augmented Generation (RAG). This technique can use more traditional software programming or even AI agents to enhance “… the accuracy and reliability of generative AI models with information fetched from specific and relevant data sources” (NVIDIA).

Indeed, these emerging AI multi-hop capabilities and techniques can radically improve supply chain decision-making. Previously, decision-makers have relied on “one-hop” static databases and manual data gathering, leading to delayed and often ill-informed decisions. With AI-powered multi-hop reasoning, businesses have the on-demand capability to access multiple data sources, analyze complex information, and make timely, informed decisions. Next, let’s look at how AI-powered Multi-Hop Reasoning can automatically break up complex problems, synthesize multiple data sources, and provide on-demand, informed recommendations to decision-makers’ toughest problems.

“Previously, decision-makers have relied on “one-hop” static databases and manual data gathering, leading to delayed and often ill-informed decisions.”

c. Examples of Multi-Hop Reasoning Across Disparate Supply Chain Systems.

To illustrate the power of multi-hop reasoning, let’s consider a few examples of complex questions that a supply chain decision-maker might ask of their business analysts. Specifically, I’ll break these questions down into sub-parts, just as a supply chain staff would need to do manually. However, with AI-powered Multi-Hop Reasoning, the system can instantly decompose these complex questions, synthesize data from multiple sources, and provide insightful answers. Without a doubt, this capability can significantly reduce the time it takes to get insightful answers, from days or weeks to on-demand. See examples of supply chain complex questions below as well as a graphic depicting single-hop question-answer versus multi-hop questions-answers.

Supply Chain Examples of Complex Questions Vs Multi-Hop Questioning

Multi-Hop Question Example 1:

• Complex Question About Delayed Shipment and Customer Satisfaction Rates:
  • “How did delayed shipments from carrier X last quarter affect our overall customer satisfaction ratings?”
• Same Question, Broken up into Multi-Hop Questions:
  • 1. “What were the delayed shipment rates for carrier X last quarter?”
  • 2. “How did these delayed shipments correlate with customer satisfaction ratings?”

Multi-Hop Question Example 2:

• Complex Question About Inventory Levels and Supply Chain Costs:
  • “What is the financial impact of maintaining high inventory levels in our warehouses on our overall supply chain costs?”
• Same Question, Broken up into Multi-Hop Questions:
  • 1. “What are the current inventory levels in our warehouses?”
  • 2. “How do these inventory levels affect our storage, handling, and maintenance costs?”

Also, it is not just current staff and supply chain systems that are challenged by multi-hop questions. Indeed, most AI Large Language Models (LLMs) lack Multi-Hop Reasoning capabilities. As a result, they also struggle with answering complex questions despite being trained on vast datasets. To address this limitation, AI developers are starting to leverage Multi-Hop Reasoning techniques. This includes Multi-Hop Question-Answer, Chain-of-Thought Reasoning, and Retrieval-Augmented Generation (RAG), often in conjunction with Knowledge Graphs. Without a doubt, these advancements enable AI-powered applications to better comprehend complex relationships and provide more accurate answers, minimizing errors and hallucinations.

For more information on AI-powered Multi-Hop Reasoning and Knowledge Graph tech, see references at the end of this article.

3. Unlocking Supply Chain Insights for Executives: A Decision System with Rapid Multi-Hop Reasoning.

I have no doubts, that we now have the technology available for executives to make on-demand, informed decisions across the supply chain. Indeed, Multi-Hop Reasoning is one of these key capabilities. Below are the technologies that make rapid, informed decision-making possible in supply chains.

Decision System Capabilities Now Available for Rapid, Informed Analytics

• Fast Computing Networks and Cheap Storage: Powers rapid AI processing, stores massive datasets, and enables swift analysis.
• Artificial Intelligence (AI): Ingests large data sets, identifies patterns, rapidly analyzes, makes recommendations, learns, and adapts. Click here for more information.
• Semantic Data Interoperability: Allows different data sources to be seamlessly integrated and understood in a common context, enabling more accurate and rapid analysis. Click here for more information.
• Knowledge Graph Tech: Enables the creation of a graphical representation of knowledge leveraging data, both structured and unstructured, that can be used to reason and infer new insights. Click here for more information.
• Multi-Hop Reasoning Methodologies: Provides the rapid ability to analyze complex questions and provide answers by reasoning across multiple data sources and hops.

Without a doubt, these decision system capabilities can provide on-demand, accurate insights for executives, helping them to make informed decisions and drive business success. Especially for corporate executives to include CEOs, COOs, and CFOs as well as supply chain leaders, they need decision systems powered by these advanced technologies to maximize supply chain performance. For more information on how supply chain executives would work directly with a decision system with these advanced capabilities, see my article, An Agile Decision Platform to Empower Executives For Superior Supply Chain Performance: Here Are The Best Attributes.

An Agile Decision Platform to Empower Executives For Superior Supply Chain Performance: Here Are The Best Attributes

More References.

• Wisecube’s article, Multi-Hop Question Answering with LLMs & Knowledge Graphs
• Moveworks’ article, What is multi-hop reasoning?
• Invisible’s article, How to teach chain of thought reasoning to your LLM
• SC Tech Insights’ article, Achieving Logistics Interoperability: The Best Way to Breakthrough The Tangle Of Dumb Data Integrations
• Also, SC Tech Insights’ article, Knowledge Graph Tech: Enabling A More Discerning Perspective For AI.
• Neo4j’s article, Knowledge Graphs & LLMs: Multi-Hop Question Answering
• IngestAI’s article, Knowledge Graphs & LLMs: Multi-Hop Question Answering – Unlocking Intricate Text Understanding

Supply Chain Analytics Types and The Way They Work To Better Empower Decision-Making

Need help with an innovative solution to make your supply chain analytics actionable? I’m Randy McClure, and I’ve spent many years solving data analytics and visibility problems. As a supply chain tech advisor, I’ve implemented hundreds of successful projects across all transportation modes, working with the data of thousands of shippers, carriers, and 3rd party logistics (3PL) providers. I specialize in launching new analytics-based strategies, proof-of-concepts and operational pilot projects using emerging technologies and methodologies. If you’re ready to supercharge your analytics or if you are a solution provider, let’s talk. To reach me, click here to access my contact form or you can find me on LinkedIn.

For more from SC Tech Insights, see the latest articles on Decision Science, Data Analytics, and Supply Chains.