Three Ways AI Makes Procurement Smarter
In 2017, Gartner predicted that artificial intelligence (AI) would benefit procurement and sourcing technology. That moment has arrived, according to Mike Quindazzi, managing director at PriceWaterhouseCoopers and top financial-tech influencer.
“We’re now in the golden age of AI, where advancements come from voluminous sets of data, new algorithms being created, computing power, and the ability to do this in the cloud at scale,” said Quindazzi.
Procurement data has exploded because procurement has evolved into something called “intelligent spend management,” which oversees all corporate purchasing processes including direct and indirect purchases, travel, and external labor.
Quindazzi cautions that while AI has many use cases in procurement, such as rating vendors, “there will always be a human at the end of AI processes, so there needs to be a sense of accountability.”
Business leaders must offer transparency into the metrics and data used, how vendors were selected and ranked, and other elements that trained AI algorithms.
At the recent SAP Ariba Live conference, SAP showcased several ways that AI and machine learning technology will impact intelligent spend management.
Consistent contract management
There are two phases of contract management during negotiations between companies and suppliers: the legal and operational. But those two phases reside in different documents and systems — and can easily become inconsistent.
Legal contracts are usually unstructured data stored in Word documents, while operational contracts are structured documents that live in either ERP or supply network applications. The operational agreements include data such as specific pricing terms, line items, or accounting information. If these details change, the two contracts can get out of sync.
For example, in the past, if the price changed for a specific item in the operational contract, it needed to be manually updated in the legal agreement. But this reconciliation didn’t always happen. New synchronization features will automatically ensure consistency across both documents.
New prototypes powered by machine learning analyze legal documents and compare them to pricing and purchasing conditions in operational agreements. The machine learning prototype can make recommendations when there are inconsistencies. For example, the solution can make sure that taxes and insurance fees in the contracts are appropriate to specific regions and countries where the transactions are based.
Chatbot-guided contract writing
Large companies generally have repositories that contain thousands of contracts. Machine learning can mine unstructured data from those repositories to help procurement professionals more intelligently write contracts and proactively avoid potential issues.
SAP is developing an AI chatbot that reviews contracts and proposes optimizations, based on historical patterns. For example, the chatbot may suggest different payment terms or recommend a particular legal clause that addresses a regionally specific insurance law.
External data is also fed into the chatbot from over 600,000 private and public sources so it knows when things like when new regionally-specific insurance laws are enacted. This enables the chatbot to flag risks and propose alternatives.
Sourcing auctions
To find the best supplier, large multinational companies hold sourcing events, called auctions. Bram Purnot, a solution architect for SAP Ariba, explained that these events help companies negotiate prices or find the best supplier based on other criteria – but holding auctions can be very complex. Some might focus on a region or a specific commodity, or a combination of factors. For example, a company may want suppliers for the cheapest, most sustainably produced shea butter in Western Africa.
To illustrate: Collaborating with a large Dutch company that holds a lot of these auctions, SAP Digital Business Services has created a prototype that uses machine learning and chatbots to help manage auctions. Once an auction has been created, the chatbots will suggest how long the auction should last, who should be invited and how many participants to include.
These chatbots are fed by a machine learning algorithm that has combed through existing data to predict how to run the most successful auction. For example, if the company is looking for transportation services, the app may recommend Japanese providers that are more effective than the Dutch auctioneer in this type of auction.
The challenge of AI and business change
AI has arrived, but organizations need help embracing new processes. Judith Hurwitz, president and CEO of Hurwitz and Associates, a technology consultancy and analyst firm, recommends that companies take a hybrid approach to determining how to use AI that includes both data scientists and procurement professionals.
“Data scientists can’t live in isolation – it’s not just an algorithm, it needs the expertise of business experts,” said Hurwitz.
Follow Robin Meyerhoff on Twitter or LinkedIn.
This article originally appeared on Forbes and is republished by permission.