The multinational manufacturer turns to cognitive computing to manage its sourcing function — and make sense of a massive amount of data relevant to supplier selection and monitoring.

Chipsets, processors and memory-storage devices manufactured by Intel Corporation made possible the modern-day age of computing, while generating a virtually infinite supply of data. Now Intel is drawing on that same wealth of intelligence to transform its own operations.

Intel’s stated goal is to transition “from a personal computer-centric business to a data-centric business.” That means getting a handle on the world of big data, the burgeoning internet of things (IoT), and sophisticated analytics to make sense of this potentially chaotic mass of information.

Intel is embracing artificial intelligence — the basis for what it terms “cognitive computing” — to manage its army of global suppliers. In the process, it hopes to make better decisions about which suppliers are the best partners for any given materials or region of the world.

The very nature of Intel’s supply chain is undergoing a profound change, driven by new processes and technologies such as hybrid manufacturing and autonomous vehicles. At the same time, the $63bn company must cope with growing volumes, increasing complexity, the impact of mergers and acquisitions, and an accompanying flood of data.

Intel works with some 19,000 suppliers to provide materials, labor and equipment required for manufacturing. But how can it tell which ones are the best for a particular component or service? Any number of supplier failures or disruptions can bring a finely tooled supply chain to an abrupt halt.

Industry 4.0 is the term given to the next wave of innovation in manufacturing, centering on such developments as IoT, cloud computing and “cognitive” technologies. That last category is the final step in a progression of analytical capabilities, beginning with descriptive and advancing to diagnostic, predictive and prescriptive systems, before achieving what forward-thinking companies view as the full realization of supply-chain automation.

Cognitive computing is expected to touch on every aspect of supply-chain management, including the Plan-Source-Make-Deliver-Return format of the Supply Chain Operations Reference (SCOR) model. But it promises to generate particular value in the world of sourcing, which up to now has been managed through a combination of limited data and human instinct.

Intel’s Objectives

Looking to raise its level of supplier intelligence, Intel set forth three primary objectives: enable commodity managers to make optimal sourcing decisions, monitor selected suppliers in an efficient manner, and continuously improve on those decisions.

The company needed accurate and real-time answers to a series of key questions. Which supplier is good at what task? Which is making progress toward developing capabilities of value to the manufacturer? Which possesses the right skills to meet Intel’s needs?

Having selected a supplier, Intel needs to monitor it on an ongoing basis, with the ability to detect organizational turmoil, financial instability or plans to acquire another supplier. As Intel puts it: “Time is critical in knowing that potential disruptions have occurred, or may be about to occur.”

Again, it all comes down to the amount, availability and quality of data. Sourcing and procurement managers must draw on information from both inside and outside the company, including research, news reports and social media. Much of that information is unstructured and not easily given to evaluation.

With the aid of cognitive computing, Intel can amass all available data, regardless of source, and assimilate it into a unified body of knowledge. The system’s natural language capabilities aid in extracting vital information from unstructured text. Using a proprietary A.I. tool known as Saffron, the company can make previously undetected connections between raw data acquired from disparate sources.

New suppliers are rated according to multiple criteria, including specific skills, job roles and geographic presence. The cognitive computing engine ranks the candidates in a series of easy-to-interpret bar graphs.

The system also creates a virtual map that reduces the need for human managers to engage in extensive research on each supplier’s capabilities. As a result, the manufacturer gains access to more relevant information, can direct requests for quotes to the most capable suppliers, acquires greater negotiating leverage when considering more than one supplier, and speeds up the entire decision-making process.

When it comes to procurement and planning, the number of variables to consider is staggering. Yet the cognitive computing platform is able to make calculations within a matter of minutes, according to the company. It can support planning across some 450 million units per year, across multiple business entities.

Managing Supplier Risk

Selection is just the first step in supplier management. Having contracted with the right vendor, Intel must then take into account the many events that can disrupt the relationship. Some, such as quality issues, might be the fault of the supplier. Others, such as natural disasters or labor disputes, lie beyond the supplier’s power to control. Either way, the result is the same: an interruption of supply, leading to higher costs and the possible loss of valued customers.

Intel’s cognitive computing platform can detect many types of disruptions by revealing connections between data inputs that would otherwise be invisible to human managers. For example, a rival manufacturer might be attempting to acquire a key supplier. Or a merger might be underway that threatens to alter the relationship with that entity.

Any such events will show up in the form of spikes within charts that are generated by the system daily. By selecting a particular point, Intel can access detailed news, social media content or other information relevant to the chosen alert.

Changes in supplier quality can also be quickly detected. The system employs “heat maps” to display any variances from key performance indicators. It can also generate “what-if” scenarios that prepare Intel to react to any quality issues that might arise.

Intel launched its supplier intelligence initiative through two pilots: outsourced product development (OPD) and corporate strategic procurement (CSP). According to the company, they resulted in a combined $30m of cost avoidance in 2017. The company expects that number to skyrocket with full implementation of the system.

Benefits from the OPD pilot alone included:

• 4-percent savings from selecting higher-quality vendors;
• 3- to 4-percent rate reduction from better negotiations, due to increased intelligence on supplier competitiveness;
• 20-percent productivity savings for each request for quotation or proposal, resulting from a one-week reduction in the time required to identify vendors for new projects; 
• Additional savings resulting from early identification of supplier insights such as mergers and acquisitions.

Looking ahead, Intel plans to extend its sourcing intelligence platform across all aspects of supplier management, including materials, labor, equipment, quality and software. Says the manufacturer: “We believe our A.I.-based sourcing intelligence will play a significant role in transforming Intel’s digital supply-chain journey.”