Over the past few years, global supply chains have faced rising complexity due to geopolitical shifts, trade barriers, and bottlenecks, which have increased the need for resiliency. In one of the recent podcasts with Emerj, Luke van der Waals, Demand-to-Deliver Value Stream Owner at SLB, explained how these challenges are compounded by inflationary pressures, making it harder to manage margins and maintain operational efficiency.
He emphasized that these trends are unlikely to reverse, highlighting the importance of building adaptable and resilient supply chain strategies to navigate an increasingly volatile environment.
AI and other modern technologies have shown significant promise in addressing the rising complexities in global supply chains. According to a study published in the International Journal of Production Research, the adoption of AI-driven tools like machine learning, predictive analytics, and digital twins enables companies to better navigate supply chain disruptions, forecast demand, and manage risks. These technologies offer enhanced decision-making capabilities, improve operational efficiencies, and foster greater supply chain resilience.
Emerj Senior Editor Matthew DeMellow recently sat with Kartik Pant, Head of Data & AI for Manufacturing & Supply Chain at Sanofi, and Shreyas Becker, Head of AI & Data Products at Sanofi, to talk in detail about these challenges in supply chain management and how to address them for improved efficiency.
Their conversation highlights the need to address supply chain challenges like demand forecasting gaps, siloed operations, and regulatory compliance while leveraging marketing insights and AI to drive accurate decisions and seamless operations.
We bring you two key insights from their conversation:
- Bridging marketing and supply chain for better demand forecasting: Align marketing signals with supply chain planning to match production with actual demand to prevent stockouts, optimize production, and boost efficiency.
- Building a supply chain control tower with a phased approach: Add predictive analytics to demand signals, inventory, and product flow to establish visibility for scenario planning and leverage generative AI (GenAI) for an interactive, decision-support system.
Guest: Kartik Pant, Head of Data & AI for Manufacturing & Supply Chain, Sanofi
Expertise: Business intelligence, artificial intelligence, data analytics
Brief Recognition: Kartik Pant is the Head of Data & AI for Manufacturing & Supply Chain at Sanofi, managing a portfolio of 15+ major applications that leverage AI to enhance operations across the company’s manufacturing and supply chain network. His work has contributed to significant operational improvements, with an estimated impact of over $200 million. He holds an MBA in Supply Chain Operations from Northeastern University.
Guest: Shreyas Becker, Head of AI & Data Products, Shreyas Becker, Sanofi
Expertise: Data analysis, machine learning, project management
Brief Recognition: Shreyas Becker is the Head of AI & Data Products at Sanofi, with a focus on Manufacturing, Quality, Supply Chain, and Procurement. He earned his MBA from the University of British Columbia, Canada. Shreyas has previously worked with Neo Financial, BCG, and PwC Canada, bringing extensive experience in leveraging AI and data to drive operational excellence.
Bridging Marketing and Supply Chain for Better Demand Forecasting
Kartik opens the podcast by explaining the interconnected relationship between marketing, supply chain, and manufacturing, emphasizing the role of marketing as a key demand signal.
He points out that accurate demand forecasting is crucial for effective supply chain planning and production scheduling. However, silos between marketing and supply chain often create a disconnect, resulting in misaligned order signals and planning.
Lack of coordination leads to challenges like stockouts and fulfillment issues, where inventory shortages and inefficiencies arise due to poor communication. He stresses the need for stronger integration to ensure smooth operations and avoid such disruptions.
Shreyas builds on Kartik’s points by positioning marketing as the core engine for demand generation. He points out that marketing not only signals demand but also directly influences production decisions by indicating “how much to make.” However, he acknowledges the complexity of interpreting marketing signals, noting that they can sometimes be misleading (“complete noises”) or genuinely indicative of actual demand.
Positioning marketing in such a way, he emphasizes, helps bridge the gap between marketing and supply chain, enhancing overall operational efficiency.
Additionally, regulatory changes, such as serialization and the growing need for serial IDs on labels, add further complexity and often necessitate last-minute label adjustments or moving packaging processes closer to the market.
Compliance-related challenges not only disrupt the supply chain but also complicate the tracking and tracing of products. Effective data management is critical to ensure compliance and maintain visibility across the supply chain for batch tracking and market delivery.
Kartik notes that, previously, these challenges were traditionally handled in a highly manual process but are now being transformed by the use of GenAII. GenAI capabilities are also being leveraged to automate the drafting and generation of compliance submissions, significantly improving efficiency while ensuring proper governance of product data.
He points out the potential of AI to streamline this process across diverse markets like Algeria, China, and Japan while maintaining a human-in-the-loop approach for oversight and quality assurance. The prospect of streamlined supply chains across the Asian Pacific and Africa for pharmaceutical players like Sanofi represents a powerful tool for enhancing regulatory compliance and operational efficiency in the pharmaceutical industry.
Shreyas expands on Kartik’s points by emphasizing two key components of demand forecasting: standard forecasting and emergent, action-based forecasting. He highlights that emergent forecasting often requires closer collaboration with marketing, as marketing plays a crucial role in responding to these dynamic situations and ensuring effective communication and adaptation to new market realities.
Building a Supply Chain Control Tower with a Phased Approach
Kartik explains a phased approach to building a supply chain “control tower,” emphasizing that it doesn’t need to be an extensive, complex system from the start:
“You can start by value stream mapping across your end-to-end supply chain. So, getting a sense of how your product actually moves in your supply chain. You can start with demand signals and try to understand, by market: What do the demand signals look like? How do they forecast or evolve? You can start with understanding your inventory, the quality of your product, and so on.
So you can build these small blocks of data and build that API that you talked about. If you connect all of these data blocks, these data assets, treat them as a product, and make sort of an overall architecture on top of it, as you’re trying to think about a control tower, end-to-end, to say, ‘What are the modules in that control tower?'”
– Kartik Pant, Head of Data & AI for Manufacturing & Supply Chain, Sanofi
The first step, Kartik says, is achieving visibility into the supply chain, including forecasts, inventory, and operational status. Only once the foundation is established, Kartik portends, can predictive capabilities be layered in, such as scenario planning and forecasting.
Kartik goes on to provide examples, such as anticipating the impact of regulatory changes in one market or addressing supplier risks upstream, to demonstrate how predictive analytics can guide decision-making.
Once this foundation is in place, traditional AI and machine learning techniques can be implemented to enhance scenario planning and forecasting, enabling the system to predict potential outcomes based on various inputs.
He then highlights the role of GenAI as an additional feature designed to improve user experience and adoption. Rather than simply providing a static, global view of supply chain data, it enables a conversational interface, allowing users to interact dynamically with the system. For example, users could ask scenario-based questions like, “What will happen if I take this action?” or “What if I don’t make this change?” GenAI acts as a “bolt-on” layer that enhances accessibility and usability, creating a more interactive, intuitive, and user-friendly experience for managing the control tower.
Shreyas builds on Karthik’s points by emphasizing that new technologies should focus on empowering users with insights and capabilities, or “superpowers,” to make more informed decisions. He predicts that the future of technology will center around creating urgency and structured processes to utilize these insights effectively.
“It could be like a chat interface in some cases. In some cases, maybe it’s building machines that talk to each other and make decisions. It could be in any of those formats. That’s how I see a lot of future technologies, perhaps because you have two systems making decisions out of two completely different pieces of information.
Then, if that is going to be a scenario, you definitely need the two machines to talk to each other; otherwise, they’ll never come to a decision. So how do you make that happen? It could be two people, in some ways. So I feel like that’s where these interfaces of the chat, or be it VR, or be it another interface – that’s how it would play out.”
– Shreyas Becker, Head of AI & Data Products, Shreyas Becker, Sanofi
