Suppose with me for a moment. Suppose you’re director of procurement at a medium sized manufacturer of control components in the Midwest. You walk into your office one bright Monday morning and say to your digital desktop assistant, “OK, “Bridget”, show me the drawdown on passive components used on assembly line 2 for the past month, and give me a 90 day order forecast to review, based on the current draw and anticipated shipments ± 10%”.

Within seconds a spreadsheet pops up on your computer screen and your digital assistant asks if you’d like to match projected component need with estimated open-market component inventories, and factor-in price fluctuations due to expected tariff hikes.

That’s the promise of artificial intelligence (AI) as it continues to change the purchasing sector. The question is, however, how far are we away from this scenario?

Since the term “artificial intelligence” (AI) was first coined in 1956 by Stanford computer scientist John McCarthy, its definition has evolved considerably. Initial usage referred to AI as “computer or machine systems that are able to perform tasks that normally would require human intelligence”.

In 1961, British scientist and mathematician Alan Turing wrote on the idea of machines able to “simulate humans and do intelligent things”. The operative word then was “simulate”.

The definition of the term, as it is used today, however, is the capacity of a computer or machine to not just “simulate”, but to actually perform operations equivalent to learning and decision making in humans.  The difference is in the addition of the word “learning”, which means machines adapting from experiences with, or as part of their environment, and adjusting their actions in response.

Machine learning really defines the new AI, which makes the use of it in procurement systems and supply networks particularly interesting. Machine learning relies on the consumption and analysis of tons of data generated by a process or system, and then learning from that data to advise on or literally control other systems.

Procurement generates lots and lots of data that AI systems can learn from, and as sophisticated sensors and sensor networks, plus the Internet of Things, become a more critical part of process control systems, inventory management systems, and transportation and shipping networks, the data sets get richer and richer.

In many procurement departments these days, however, analytics and big data rule and data quality and integration take a back seat. This means that relatively small cost containment gains from volume pricing, batched delivery, or other purchasing efficiencies take precedence over things that AI is good at, like identifying cognitive bias in procurement, which could lead to much larger efficiencies.  

In fact, AI may have the greatest immediate impact in alerting companies about factors affecting their performance that have not been considered. Having an understanding of an organization’s total spend, through spend analytics software, can make sense of volumes of spend data and lead to consolidated spending, reduced costs, faster processes and greater overall efficiency.   

Use of AI in the purchasing function, and with corporate commitment to integrated data collection and smart processes, can also lead to opportunities with overall strategic thinking and sourcing. The development of cognitive procurement advisors (CPAs) and virtual personal assistants (VPAs) using natural language processing, as with “Bridget” in our opening example, can lead to increased automation and efficiency throughout the procurement function.

For example, CPAs can produce summaries, give advice and make recommendations affecting supplier assessments, performance, risk management, and compliance. Software supplier SAP Ariba already offers an AI-driven package that can understand company purchasing policies and procedures and implement them in the invoicing and payments area to reduce errors and speed processing.

The short story in all of this is that AI systems either already exist, or are being developed that promise capabilities that include, but are not limited to:

• Detection of incomplete or inaccurate internal spend data
• Automation of mundane tasks (negotiating with suppliers, updating inventory lists, etc.)
• Comparison of internal data with external trends and individual supplier data to better manage risk
• Tracking and reporting on exchange rate volatility and its effect on spend
• Automated purchasing suggestions based on historical trends
• Insights on internal vs. external procurement function performance and spend patterns
• Vendor recommendation matched to specific production needs
• Identification of opportunities for quicker processes, reduced costs and greater efficiency
• Analyzing external demand trends to predict pricing patterns

The availability of this analytical processing power, however, does not mean that wide-scale adoption of it is happening apace. In fact, a recent study by the Hackett Group showed that even though 85% of corporate executives at large companies believe that digital technology will change the way they deliver services over the next 3-5 years, only 32% of them have implemented a digital strategy in their company.

Despite this overall digital strategy vacuum however, Forrester Research notes that 55% of companies will be investing in some form of AI before the end of 2019, and Gartner Inc. research says nearly 50% of large global companies will be using AI by 2023 in supply chain operations.

Clearly, the reality of some form of AI-augmented software usage across industry, and in the procurement function, is beginning to catch up to the promise. If you work at a large, global company, chances are you’re already knee-deep in implementing various strategies that employ AI. In fact, you may already be implementing AI-assisted robotic technology for order fulfillment and discussing the increased use of AI driven robots to assist warehouse workers.

Technology development and practical adoption do not always go hand-in-hand. But technology development in the early part of the 21^st century, particularly the use of AI in smart procurement systems, is moving faster, fast. Transactional work is becoming automated. Decisions based on the analysis of vast metrics are soon to be common. Predictions based on the same metrics will certainly follow.

All this means that the procurement function in organizations, both large and small, will increasingly move from operational to strategic. AI will have a large effect on the speed of this change. One thing is for certain. AI is here to stay.