AI-powered demand forecasting for accurate inventory planning and proactive decision-making in
logistics.
Consider this: it takes an average of 6 months to manufacture a tire (tire design, raw material sourcing, mixing and extrusion, tire building, curing, quality assurance, and packaging) and during this time, the market can undergo significant changes. Without precise forecasting, businesses risk being unprepared for fluctuations in demand, leading to excess inventory or stock-outs.
I recently attended an event CII IL ‘Solution de Technology, Automation & Robotics‘ on 21 July 2023, Hotel The Lalit, New Delhi. Session Chairman Mr Rishi Diwan did a great job of picking great panelists. In a thought-provoking panel discussion, top industry leaders from the logistics sector gathered to shed light on the game-changing impact of automation. The distinguished panelists included Mr. Anil Syal, President of Safexpress Private Limited; Mr. Devang Mankodi, Vice President – Shared Service at DP World; Mr. Manish Kumar Gupta, Head – Spare Parts Logistics, Honda Cars India Ltd; Mr. Rajesh Gupta, Head-Supply Chain Management, JK Tyre & Industries Limited; and Mr. Vikas Kalra, Head-Supply Chain Management, Hindware Home Innovation Limited.
The logistics industry is undergoing a profound transformation with the advent of automation. This revolutionary technology is redefining operations, fostering efficiency, and propelling growth. Embracing automation empowers logistics providers to streamline processes, boost productivity, and expand their business horizons. By harnessing the full potential of automation, these companies can attain operational excellence, unearth inventive solutions, and maintain a competitive edge in a constantly evolving market.
Demand planning and forecasting is the weakest link in the chain
The bear game is a popular logistics training exercise where participants simulate managing the movement of goods in a supply chain. It demonstrates the importance of demand forecasting in logistics by highlighting how accurate predictions help participants anticipate and respond to fluctuations in demand, allowing them to optimize inventory levels and ensure efficient supply chain management in a dynamic environment.Each participant takes on the role of a different supply chain stakeholder, such as a manufacturer, distributor, or retailer. As the game progresses, participants face challenges like transportation delays, stockouts, or sudden changes in demand.
Traditional forecasting vs AI-powered forecasting
While traditional statistical demand forecasting relies on historical data spanning 24 to 26 months, its limited scope and inability to capture real-time factors hinder its accuracy (60/75%). In contrast, AI-powered demand forecasting, leveraging real-time data and machine learning algorithms, can surpass traditional methods. By incorporating variables like GDP growth, market dynamics, weather conditions, and more, AI-driven models can achieve 95 to 98% accuracy, empowering businesses to make data-driven decisions and stay agile in the face of market changes.
Similar to how Google Maps dynamically updates its estimated arrival time based on real-time traffic conditions, AI-powered demand forecasting should continuously adapt to changing market dynamics.
How generative AI can make forecasting for logistics more accurate
Generative AI, specifically Language Models like Large Language Models (LLMs), holds great promise in revolutionizing demand forecasting. LLMs can analyze vast amounts of textual data from diverse sources, such as customer reviews, social media, news articles, and industry reports. By comprehensively understanding customer sentiments, market trends, and emerging events, LLMs provide valuable insights for demand forecasting models.
Scenario Analysis made possible by generative AI
LLMs can generate hypothetical scenarios based on given conditions, helping businesses anticipate various demand outcomes. This scenario analysis allows demand planners to explore “what-if” situations and strategize effectively for different scenarios.
This is how it could work:
AI Assistant: (Responding) Sure, let’s get started. Please upload the historical sales data and any other relevant market information.
User: (Uploads data) Here is the data for the past 24 months, including sales figures, customer feedback, and economic indicators.
AI Assistant: (After analysis) Customer sentiment has been positive over the past few months, with a notable increase in interest for our new product line.
User: That’s promising! What about upcoming market trends?
AI Assistant: Predicts a potential surge in demand for our products due to a major industry event scheduled next month. Additionally, the analysis indicates a potential slowdown in customer interest during the following quarter.
User: Interesting! Based on these insights, what’s the forecast for the next quarter?
AI Assistant: With the current positive trends and the anticipated industry event, the LLM forecasts a 15% increase in demand for the next quarter. However, it also suggests being cautious about the following quarter due to the potential slowdown.
LLMs being used to intelligently collect insights for accurate forecasting
LLM can engage with customer service representatives, product managers, and sales teams to extract valuable insights from their interactions with customers.
Customer Service Representative: Hi! Sure, I can help. We’ve been receiving a lot of inquiries about the new clothing line, especially about the availability of certain sizes and colors.
AI Assistant (LLM): Thank you for sharing that. Are there any recurring patterns in customer inquiries that you’ve noticed?
Customer Service Representative: Yes, customers have been asking about the fabric composition and the environmental certifications of our eco-friendly products.
AI Assistant (LLM): That’s insightful. Do customers frequently provide feedback about the shopping experience related to our sustainable fashion line?
Customer Service Representative: Yes, we often receive positive feedback about the quality and comfort of our eco-friendly clothing, and customers appreciate our efforts to promote sustainable fashion.
AI Assistant (LLM): (Initiating chat with Sales Representative at GreenEco) Hello! I’m the LLM-powered demand forecasting assistant. I’m interested in learning more about sales trends and customer preferences related to our new eco-friendly clothing line.
Sales Representative: Hi! I’d be happy to help. The sales for our eco-friendly clothing line have been steadily increasing, especially for our organic cotton t-shirts and recycled polyester jackets.
AI Assistant (LLM): That’s insightful. Can you identify any specific factors contributing to the increased demand for these products?
Sales Representative: Yes, customers are increasingly conscious of sustainable fashion choices, and they appreciate the versatility and style of our eco-friendly clothing.
AI Assistant (LLM): Thank you for sharing that. Have there been any notable changes in customer preferences or market trends that you’ve observed?
Sales Representative: We’ve noticed a growing interest in earth-toned colors and minimalist designs, aligning with the current sustainable fashion trends.
AI Assistant (LLM): Valuable information! Now, let’s integrate these insights into our demand forecasting process for the new eco-friendly clothing line.
AI Assistant (LLM): (Analyzing customer service and sales data) Based on the chat with customer service, there’s a growing interest in our new eco-friendly clothing line, particularly regarding size availability, fabric composition, and environmental certifications.
Conclusion
From the conversation with the sales team, we can observe an increasing demand for our organic cotton t-shirts and recycled polyester jackets. Customers’ conscious choices towards sustainable fashion, along with the appeal of earth-toned colors and minimalist designs, are driving this trend.
By integrating these insights into our demand forecasting model, we can anticipate continued growth in demand for our eco-friendly clothing line. We can focus on optimizing inventory for popular products like organic cotton t-shirts and recycled polyester jackets while considering customers’ preferences for fabric composition and environmental sustainability.