For the past 50 years, businesses have been digitalizing their processes to save costs and improve efficiency. In the latest iteration, algorithms based on human intelligence are being used to analyze large amounts of data to reveal anomalies, detect patterns, and identify trends.
Collectively these algorithms are known as artificial intelligence. AI has already changed the economics of certain industries such as the automotive business with algorithms that give cars the ability to drive themselves. Those same algorithms are poised to change procurement as sweepingly by creating purchasing systems that make buying decisions by themselves.
AI is a big term covering many separate technologies including rules-based expert systems, machine learning, deep learning, machine vision, natural language processing and General AI. Although General AI has captured the popular imagination with a vision of a single entity having similar or superior overall intelligence to human beings, Narrow AI, the application of specialized algorithms to limited problems, is likely to play a more prominent role in digitalizing the procurement process to create a self-buying purchasing system.
Areas in procurement where AI plays a role
The process of finding, negotiating terms and acquiring goods (or services) from an external source, often using competitive bidding has become increasingly complex. To ensure the buyer receives the best possible goods, works, or services at the best possible price, many aspects of products such as location, quality, time and quantity must be weighed and determined. Almost every purchasing decision also includes external factors that must be taken into account such as price changes, marginal benefit, delivery and handling. Purchasing also requires making buying decisions under conditions of scarcity.
Increasingly it is considered industry best practice to make use of empirical statistical approaches such as cost-utility or cost-benefit analysis. Governments and industry associations often mandate today how purchasing will be handled to encourage open and fair competition while minimizing collusion and fraud. Add to that the unprecedented explosion of gigabytes and gigabytes of data available to purchasing departments and you have a situation that is beyond the capability of human reason to handle.
Just as automobiles can navigate and steer themselves, shortly purchasing systems will have the ability to identify and obtain goods, services and works by themselves. They will be able to adapt their purchasing decisions to changes in price, benefit and delivery using machine learning. They will be able to analyze gigabytes and gigabytes of data to identify patterns of collusion and fraud using deep learning. They will be able to demonstrate compliance with governments and industry association requirements using rules-based expert systems.
It is unlikely that a single purchasing robot algorithm will be able to perform all of this. Just as a self-driving car depends at a high level on independent computer vision, sensor fusion, localization, path planning and control systems, self-buying purchasing systems will likely depend on independent algorithms to perform each of the following purchasing tasks:
- Transaction detail analysis.
- Inventory management.
- Consumption and usage data analysis.
- Contract terms and rates rules identification.
- Market information analysis.
- Historical pricing analysis.
- Issue resolution analysis.
Unlike the self-driving car, the self-buying purchasing system will not need all of these algorithms to be integrated to start to add value. Instead, they are likely to be introduced piecemeal. The first iteration of AI will be computer aided purchasing where one-by-one these systems provide inputs to human procurement agents who still make the final buying decisions.
Benefits of AI in procurement
As they eventually become integrated, these systems will assume more and more of the responsibilities of procurement until finally, they are making all the buying decisions by themselves. AI systems are not perfect, any more than any computer system (and think about how many times you have rebooted your personal computer) so any self-buying system will require management processes to ensure no errors. These will still have to be managed by human beings. Human beings will also have to monitor, expand, and maintain the computer hardware these algorithms need. In the future, the role of purchasing professionals will change, from making all purchasing decisions based on their experience and judgment to making decisions partly based on algorithmic inputs, to not making buying decisions other than typical ones.
Just as the self-driving car promises to increase the safety, efficiency and capacity of transportation for the better, AI-based self-buying systems will lower the cost of procurement, buying goods at higher speeds, with more visibility and offering capability beyond what we can imagine.
The incremental improvements in AI-based procurement processes are most likely to be introduced by cloud-based spend management software companies like Negotiatus that already help companies centralize and streamline their purchasing process.
The competitive advantage AI in procurement systems will provide early adopters is likely to be staggering.