In the ever-evolving landscape of business functions, GenAI (Generative Artificial Intelligence) has emerged as a transformative force. From streamlining operations to enhancing decision-making processes, GenAI is making waves across various sectors. Procurement, a critical aspect of business operations, is not immune to this wave of innovation.  

According to an article by KPMG, GenAI can help assist and augment all procurement functions, such as category management, strategic sourcing, contract lifecycle management, supplier risk and relationship management, requisition to procure items, invoice-to-pay, and so on.  

In this blog, we are delving into the intriguing realm of leveraging GenAI in spend classification and exploring some significant benefits it brings to the procurement table. 

Before we explore the role of GenAI, let us understand the concept of spend classification. In procurement, spend classification function comprises categorizing spends into distinct groups, aiding organizations in analyzing and managing their budgets effectively. Classified spend data helps in bringing visibility into how effectively the dollars of the organization are being utilized. It helps procurement professionals understand which are the high-spend buckets and provides opportunities to consolidate spends and select more reliable suppliers on further analysis.  

Typically, spend transaction data might be expressed with short-forms or non-standard abbreviations. In fact, two spend transactions of the same item could be expressed differently from one another and that makes simply classifying this data much more challenging. 

What makes it even more complex to classify spend manually, is that millions of data lines of spend transactions need to be bucketed by an individual, which is a long and tedious process and prone to errors. The efficiency and accuracy of a traditional spend classification depends on the cognitive ability of the human classifying the spend and the results could be non-standard and even grossly inaccurate.  

There are other rule-based methods of spend classification, but the accuracy level of these approaches is quite low and seldom add much value when spend buckets are being analyzed. 

ignio™ Cognitive Procurement, an AI-powered spend intelligence software, offers AI-based spend classification, which is leaps ahead of the traditional methods. It eliminates the two main issues that conventional classification function endures. First, it is much faster than the manual approach. Thousands of line items can be classified in a few hours as compared to a few weeks using the traditional method. Furthermore, the accuracy of classification is at least 3x higher than the traditional method, as it follows a standard approach and does not depend on an individual classifying the data. 

To further improve the accuracy of spend classification, ignio™ Cognitive Procurement has an ‘Ask GenAI’ functionality, which has revolutionized this landscape, offering a more efficient and accurate approach. 

GenAI’s prowess lies in its ability to generate more information about the prompt using the AI models that are trained to recognize the patterns in the prompt.  In the context of spend classification, this translates to finding more information about the spend data line item by analyzing the abbreviation, understanding the context, recognizing an image related to the spend data. This information helps enhance information related to the transaction to be classified. Using this supporting information generated by GenAI, it is possible to ameliorate the accuracy of spend classification. It also helps in reducing the burden on procurement professionals, and thereby minimizing the risk of human error. 

ignio™ Cognitive Procurement uses the state-of-the-art AI (Artificial Intelligence) to offer spend insights and recommendations. ignio™ Cognitive Procurement starts the process with classifying spend transaction data according to industry-recognized or custom taxonomies.  

The process of spend classification starts with selecting the purchase orders transactions that need to be classified. Once the data set is defined, the spend data is classified using the taxonomy defined. There are numerous types of industry-recognized taxonomies, such as UNSPSC, SIC, NAICS, NACE, and so on. However, organizations may choose a custom taxonomy, and ignio™ Cognitive Procurement can be trained to classify using those nomenclatures. 

Post AI-based classification, the output shall be hinged with the machine learning-generated confidence values. A high confidence value shows that the classification has a high degree of accuracy. But sometimes, the classification might be hinged with a low confidence value. This may be due to a close tie between two categories where a certain line item can be bucketed.  

A low confidence value line item can be assigned to a human for verification or reconciliation. But this may expose to opportunity for errors, as the classification depends on the judgement of a human. Moreover, it can be a time-consuming process. And overall, these challenges can lead to reduced accuracy of data classification. 

Enter GenAI

Where the results of AI-based spend classification falls short, there the GenAI-based spend classification steps in to augment the categorization and enable the enterprise a greater chance of taking better procurement decisions owning to the high degree of visibility in their spend. 

ignio™ Cognitive Procurement’s spend classification module enables the use of GenAI to increase the accuracy of classified data.  

Let us take an example of a line item that needs to be classified as follows: 

‘File’ 

When we manually type in such a line-item text in Google Search, it could display results on both categories – a folder to organize papers or a tool to smoothen sharp surfaces. Now, if we assign this to a human to classify, it can be quite confusing for them as an enterprise could use both, but the first file would be an administrative expense while other would be a manufacturing expense. Having such numerous cases could misrepresent the spend buckets and drive procurement professionals to take wrong decisions based on this manipulated information. 

We use GenAI as an augmentative application, which gives us a targeted response. It enriches the information related to the line item by providing additional information on factors boosting efficient decision-making, such as usage, application, composition and brand. 

By processing such information, it can classify the line item ‘file’ as a stationery item or tool depending on the supplier’s name, brand name and generate further information based on the data it can extract from the transaction line item. 

This classified data is fed back to further train the AI models, thereby increasing its accuracy by 5-10%, and into the Spend Analysis module of ignio™ Cognitive Procurement to produce spend insights and out-of-the-box recommendations, which can be immediately actioned by the procurement professionals. These insights reduce the time consumed in the procurement decision-making process. Using these insights, the processes can become more resilient, compliant and result in greater savings. 

In conclusion, integration of GenAI in spend classification is a game-changer for procurement functions. Automated data analysis, enhanced generative capabilities, and optimized real-time decision support, offered by GenAI, redefine how organizations manage their spends.