AI in DDMRP - Innovations for the Future of Supply Chain Management

The Vision

The dawn of artificial intelligence (AI) in supply chain management is not just on the horizon—it's here. At b2wise, we've already developed and deployed machine learning forecasts, segmentation, and buffer optimization solutions. This year, we formed an Innovations Team, aiming to harness the power of AI 'To make the life of our planners easier.' Our vision is to leverage the vast amounts of data from our 170 clients and use it to develop powerful AI bots that can be easily tested, enhanced, and securely deployed across all our clients.

During a recent workshop in Cape Town with an esteemed AI expert, from a new client that has heavily invested in AI, and my newly formed innovations team, I gained some valuable insights into how and what to develop and deploy. In this newsletter article, I aim to give you a glimpse into the key steps we discussed to get this exciting development underway.

Step 1: Data

Everything in AI begins with data. Our AI expert emphasized the need for a robust data foundation, noting that all AI initiatives require large volumes of continuous data and most projects stumble because gathering this data is hard and time-consuming. Fortunately, our innovations team has already mastered this issue with the use of our Datamart, which normalizes and stores millions of rows of relevant supply chain data after each daily run of our planning application. This vital first step sets the stage for the sophisticated work that follows.

Final Step: Deployment and Visualization

Growing up in a world where connecting the dots was a methodical process, I was intrigued when the discussion fast-forwarded to deploying AI bots and visualizing data through Business Intelligence (BI) tools. It seems that with AI, your 2 major issues are data inputs and data outputs. There seem to be enough skilled resources and tools around to build the AI bots that provide the insights.

As it turns out, the most popular AI data platforms are AWS and Azure. At b2wise, we have just switched from Azure to AWS to leverage their serverless architecture, which enables us to process hundreds of thousands of SKUs within hours. With AWS and their use of secure S3 buckets, it's surprisingly straightforward for AI developers to work with the data securely and then write back the results into our DataMart for interpretation and visualization.

The Middle 2 Steps: Imagining and Building the Model

But what of the middle steps? What exactly do we want AI to accomplish? It's here where industry experience and creativity meet technology. We're not interested in another forecasting predictive AI model; we are seeking to push boundaries and deliver something refreshingly different. So, let me outline the transformative areas our Innovation teams have started working on with a select few customers:

• Dynamic Buffer Adjustment: We're leveraging AI's analytical prowess to fine-tune buffer sizes, reducing carrying costs and preventing stockouts.
• Inventory Network Rebalancing: For clients with large networks, we are adding transportation costs to our model to suggest redeployment opportunities before repurchase.
• Supplier Performance Analysis: Using up to 3 years of supply data, we are looking to dynamically adjust lead times over the year and even suggest alternative near-sourcing options to cater to supply disruptions.
• Predictive Analytics for Risk Management: By anticipating disruptions, AI allows us to run rapid scenarios aimed at identifying our weak spots and allowing us to fortify our defences before the challenges arise.
• Process Optimization: In b2wise, we've built-in process orchestration, so we will use AI to identify inefficiencies in our process so that we minimize the addition of internal administration time to the placing of orders or fulfilling customer demand.
• Innovative Use of Large Language Models (LLMs): The incorporation of LLMs like GPT-4 allows us to interpret user notes and e-mails so that we can quickly summarise large volumes of data and gain insights from across the organization to enhance decision-making.

Conclusion

AI is coming, and it offers such wonderful opportunities. I believe that DDMRP as a methodology, with its built-in continuous feedback loops which require us to store all of your supply chain data daily, is extremely well-suited to harness the power of AI.

This time last year, our focus was on processing speed, which we have accomplished by rewriting our entire backend. If we did this in one year, I cannot wait to see where we will be with our new UI and AI one year from now.

If you have a wonderful supply chain AI idea, I would love to hear it.

#ddmrp #ddbrix b2wise Demand Driven Institute Association for Supply Chain Management (ASCM)