This is a great new era for JDA and our customers. We’ve been making the hard turn toward adopting machine learning (ML) across our portfolio. With the announcement of a definitive agreement to acquire Blue Yonder, we have taken our boldest and most impactful move in this transition. Blue Yonder is exciting because it helps JDA unleash tremendous value locked in the data sets of our customers by delivering the first productized artificial intelligence (AI) solution in the supply chain marketplace.
Why? Because AI has started to transcend human computational intelligence across many fields.
AI is a foundational technology paradigm that dramatically alters how software is developed and used. Planning software has long been designed to work on steady state datasets. Steady state is often achieved by aggregating nodes—such as stores, consumers and products—until they can be modeled with a single behavior or, in mathematical terms, can fit a single distribution function. This way, billions of independent ordering, ranging, expediting, allocating and promising decisions needed in an extended supply chain are replaced by a few thousand daily or weekly batched decisions that can fit the computational limitations of traditional technology stacks and be understood by a small team of operators looking at “top 10 issues” dashboards.
In retail, while traditional statistical methods have improved forecast accuracy at aggregate levels, predicting customer demand by store, by SKU every day has largely remained ineffective. It is common to see hundreds of store operators exercise their judgment on key decisions such as how much to order or what localized pricing to set. As a result, retail stores oscillate between out-of-stock situations and waste, largely at the mercy of an operator’s ability to assess factors such as weather and local events. An intelligent agent that can predict daily demand at every store for every SKU, taking hundreds of internal and external signals into account, can replace repeated manual tasks such as ordering and pricing with tremendous accuracy. This leads to millions of dollars in savings through increased productivity, waste avoidance and strategic pricing.
In manufacturing, agents can predict how much inventory needs to be carried every day at a product or location level, allowing companies to attain desired levels of service consistently while shaving out unnecessary inventory at multiple echelons of the supply chain. Static lead times in transportation networks can be replaced by dynamic lead times that look at current traffic conditions, weather and port congestions, and can predict exact ETA, reducing unnecessary expedite costs while driving customer satisfaction. In specific use cases such as spare parts supply chains, this increased predictability can reduce inventory costs by tens of millions of dollars annually.
ML-based systems, with their ability to model every node based on its own behavior, fundamentally challenge the need for steady state. By treating each store, SKU or even consumer as a distinct distribution function, these systems can:
- drive optimal assortment and ranging by product at each store
- drive lowest inventory levels to meet desired service level agreement (SLA)
- drive dynamic pricing in the ecommerce world by modeling real time competitive elasticity, and
- predict failures at the factory using real time machine data.
These systems make billions of decisions every day, unlocking value around unnecessary inventory, labor, waste and expedite charges, while driving top line with better availability and pricing. They do all of this with a scale and quality well beyond the reach of current systems and their users.
So, this brings me back to Blue Yonder and its differentiated value for solving all the problems above. With Blue Yonder, JDA is not building a better mouse trap, but fundamentally altering the value realization at our customers. Layered on top of JDA’s end-to-end supply chain solutions, the new AI-based solutions will add a new dimension of dynamic decision-making. By successfully productizing the core elements—a proven technology stack in the cloud, domain specific model creation, automated variable selection and external data ingestion—the Blue Yonder solution eliminates significant implementation risk inherent in bespoke machine learning solutions at our customers. At the same time, by offering Blue Yonder on our extensible JDA Luminate platform, the JDA approach allows customers to retain many proprietary algorithms in-house, while still leveraging the differentiated capabilities Blue Yonder brings.
We are fast approaching singularity—the state in which machine intelligence overtakes human intelligence in core supply chain solutions. The combination of JDA and Blue Yonder will now make it available to our customers.