Executive Summary

Presenting Rahul Vishwakarma from Crest, our portfolio company in our latest Expert Corner session. The impact of AI on supply chain execution, including agentic supply chain.

Rahul challenges a commonly held misconception: supply chain isn’t just a logistics issue: it’s fundamentally a capital efficiency problem. 

He also makes a compelling case for decision-making in the AI era, where speed consistently outperforms precision. An 80% accurate forecast acted upon instantly beats a perfect one delayed by days. At Crest, they are building a digital twin of sorts ie. a “living control tower” that is powered by AI agents that continuously monitor inventory, demand signals, and external factors; enabling near real-time decisions without human bottlenecks. The result: dramatically improved inventory turns and significant reductions in working capital. 

ch the full session to dive deeper into these insights and stay tuned, as we explore more on the AImension later in ts newsletter.

 1. The Problem: The Complexity of Global Networks

Crest defines the modern supply chain not as a simple line, but as a high-dimensional optimization problem.

• The "Curse of Dimensionality": With thousands of SKUs, hundreds of suppliers, and global shipping lanes, the number of possible configurations for a supply chain exceeds the number of atoms in the known universe.

• The Goal: To find the "Global Minimum" for cost while maintaining a "Local Maximum" for service levels and resilience.

2. Moving from Heuristics to Mathematical Programming

A major portion of the video critiques "traditional" supply chain management, which relies on human intuition and simple spreadsheets.

• The Shift: Rahul advocates for Linear Programming (LP) and Mixed-Integer Programming (MIP). These are mathematical frameworks that allow a computer to evaluate every possible move (where to ship from, which port to use, how much inventory to hold) simultaneously.

• Shadow Pricing: A key technical takeaway is the use of "Shadow Prices" to understand the marginal cost of constraints; essentially asking, "How much would I save if I increased my warehouse capacity by just 1%?

3. Digital Twins and Simulation

The video emphasizes that you cannot optimize what you cannot simulate.

• The Model: Building a "Digital Twin" of the supply chain allows companies to run "What If" scenarios.

• Stress Testing: Rahul highlights using these models to simulate "Black Swan" events (like a port closure or a sudden spike in raw material costs) to see where the chain breaks before it actually happens.

4. Demand Sensing vs. Demand Forecasting

Rahul makes a vital distinction between traditional forecasting (looking at the past) and Demand Sensing (looking at real-time data).

• Latency: Traditional supply chains suffer from "The Bullwhip Effect," where small changes in consumer demand cause massive fluctuations further up the chain.

• AI’s Role: AI is used here to ingest non-traditional data (weather, social media trends, port congestion data) to adjust inventory levels in real-time, reducing the "safety stock" that usually sits idle and wastes capital.

5. Multi-Echelon Inventory Optimization (MEIO)

This is a core technical pillar of the video. Instead of looking at a single warehouse in isolation, MEIO looks at the entire network as one organism.

• The Optimization: It calculates the exact right amount of inventory to hold at every stage—from raw materials to regional distribution centers—to ensure that the final customer gets their product with the least amount of total system capital locked up.

6. Sustainability as an Optimization Constraint

Finally, Rahul argues that CO2 reduction is no longer a "Nice to have" but a mathematical constraint.

• The Trade-off: By optimizing for distance and full truckloads (to save money), you often inherently optimize for lower carbon emissions. Rahul argues that "Efficiency is Sustainability."

To dive deeper into these insights and hear the full discussion, watch the complete webinar video.