Share this article
Follow us on Linkedin
Artificial intelligence (AI) technologies are playing an increasingly significant role in transforming the way businesses operate worldwide. In 2023, there was a true revolution associated with the availability of AI platforms, paving the way for new possibilities and innovations. The new functionalities and AI advancements have initiated changes that are currently reshaping the business landscape, with one area that has significantly gained importance being purchasing management.
In this context, the application of AI in purchasing is particularly intriguing and significant, as it has the potential to revolutionize supply chain management and procurement processes, enabling companies to achieve higher levels of efficiency, cost optimization, and better control over business operations.
“Before you embrace AI in purchasing”
Before diving into the embrace of AI in purchasing, however, a crucial step is to digitize our process Source to Pay. This is foundational, without which any AI efforts may prove ineffective or even counterproductive. Just as the Facebook application needs access to the internet to provide users with interaction and platform usage, AI similarly requires a solid digital foundation to function effectively. The digitized Source to Pay process provides the necessary data, forming the basis for AI operation. Consequently, AI can analyse historical purchasing data, forecast trends, identify optimization areas, and automate routine tasks. Without this foundation, AI lacks the context and information needed to make informed decisions.
Additionally, the digital tools used in the process ensure a consistent flow of data, which is a key element for the effective operation of artificial intelligence. Through the digitization of the purchasing process, information is gathered in one place, facilitating both data collection and processing, enabling effective analysis by AI-based systems.
Therefore, investments in digitizing the Source to Pay process are a crucial step before implementing AI in purchasing. Such a strategy allows companies to build a solid digital foundation necessary to harness the full potential of artificial intelligence in procurement processes. Without it, AI may only be an ineffective tool that fails to deliver expected benefits and does not support effective management of corporate expenditures.
Where is the line between now and then?
There are many different AI models, each with its own applications, methods of operation, and levels of advancement. These diverse forms of AI offer a wide range of capabilities and potential benefits for businesses.
There are many different AI models, each with its own applications, methods of operation, and levels of advancement. These diverse forms of AI offer a wide range of capabilities and potential benefits for businesses.
It is not my intention to bore the reader by listing and describing each of these models, but it is worth noting that they are often used in various areas of machine learning and data analysis. For example:
Deep Neural Networks (DNNs) are often used in image recognition and natural language processing.
Logistic Regression is popular in binary classification (e.g.spam detection).
- Logistic Regression can be used to predict numerical values (e.g., forecasting demand for a product based on planned changes in prices, promotions, etc.).
- Decision Trees along with Random Forests are often used in classification problems (e.g. identifying potential suppliers based on various criteria such as product quality, prices, delivery timeliness, financial stability, etc.).
- The Nearest Neighbour Algorithm – finds application in spatial data analysis.
- “Naive Bayes” is often used in classification tasks, especially in the context of text analysis.
The aim of this article is to outline a typology of AI levels, which will allow you to better understand and plan its use for business purposes, particularly in support of purchasing. For the purposes of this approach, I will present the division of artificial intelligence into weak and strong levels, where:
- Weak Artificial Intelligence (ANI, Weak AI): Encompasses systems that can perform specific tasks or functions according to programmed rules or algorithms. This includes machine learning models, including statistical modelling, neural networks, decision trees, etc. These models learn from data, and their goal is to perform specific tasks such as classification, forecasting, or data analysis.
- Strong Artificial Intelligence (AGI, Strong AI): This concept refers to a theoretical form of AI that would have self-awareness and the ability to think generally, analogous to the human mind. Such models do not currently exist and are the subject of discussion in the fields of philosophy and science. These are not specifically defined AI models but a theoretical concept that surpasses current achievements in the field of artificial intelligence.
It is worth noting at this point that in further considerations, I will focus on the practical application of weak-level AI, tempering somewhat the expectations of many business leaders regarding the level of AI support not only in purchasing but also in conducting all key processes in the enterprise.
New and old sources
Within the weak-level AI, we will operate within two types of artificial intelligence: generative AI and discriminative AI, where:
- Generative AI focuses on generating new data or content based on the analysis of existing data. Generative models can create new examples based on patterns and probability distributions and can be used for predicting purchasing trends or creating new category purchasing strategies, while.
- Discriminative AI focuses on classification and analysis of existing data, helping to identify patterns, trends, and relationships between them. Discriminative models are used to distinguish and classify data based on previously processed examples. They are used for fraud detection, anomaly detection, procurement process optimization, real-time category purchasing analysis, segmentation, supplier identification, risk assessment, or automation of the offer evaluation process.
Why am I mentioning this classification? Personally, I argue that the combination of both types of artificial intelligence can bring additional benefits by complementing each other and providing a comprehensive approach to purchasing management. Thus, companies can leverage the full potential of available weak-level artificial intelligence (ANI) for optimizing their procurement processes, increasing efficiency, and achieving better business results.
Practical application of AI here and now
Sourcing – Within the SAP Guided Sourcing system, the RFX process is supported by AI. By utilizing artificial intelligence, the buyer can swiftly implement the requirement, propose suitable suppliers to address the RFX, using historical data, evaluations. AI in SAP Guided Sourcing identifies qualified suppliers and provides valuable insights for informed decision-making by the buyer. A comparison of the process “without” and “with” the use of AI is presented in Table 1.
Table 1
| No-AI support | AI support | |
|---|---|---|
| Buyer | Buyer | AI |
| Initiation of a sourcing event | Initiation of a sourcing event | |
| Manually importing lines of Purchase Requisition | Intelligent Data Import & Extraction (XLS) | |
| Manual review of supporting documentation | Supplier List Recommendations | |
| Manual selection of questions/requirements for suppliers | Verification of recommendations and selection of suppliers | |
| Manual selection of suppliers | Prerequisite recommendation for vendors | |
| Publishing the RFX | Defining the prerequisites for the event based on recommendations | |
| Sending RFX to suppliers | Supplier selection support | |
| Publishing the RFX | ||
Purchase Orders– Within the SAP One Spend Guided Buying system, the purchase process has been simplified and expedited, for example, engineering orders. The system allows an employee (not a buyer) to describe their needs in natural language, and AI, deducing their intentions, allows for the rapid interpretation of the need and translation into specific products or services. A comparison of the process “without” and “with” the use of AI is presented in Table 2.
Table 2
| No-AI support | AI support | |
|---|---|---|
| Buyer | Buyer | AI |
| The need to order a product or service | The need to order a product or service | |
| Manually searching for a suitable proposal in the purchasing system | Introducing an order description in natural language | AI support in the purchase system's understanding of natural language phrases |
| Manual identification of related products and services | Recommendation of products and services that match each other | |
| Order | Selection and order | |
| Indication and information about the status of the order | ||
Category Management – Within the SAP Ariba Category Management system, the category management process has been automated, which, in addition to defining category profiles and creating procurement strategies based on them, allows for integration with source-to-pay systems. This enables AI support in identifying and transforming initiatives into opportunities as projects or sourcing events.
These are just a few examples of already available functionalities supported by artificial intelligence in SAP ARIBA systems for effective purchasing. Soon, we can expect new applications. Personally, I expect:
- Intelligent generation of catalogue item descriptions for suppliers: This functionality will allow for automatic generation of catalogue item descriptions based on available data, making it easier for suppliers to create and update product catalogues. This will make the catalogue management process more efficient and accurate, potentially speeding up purchasing processes and improving user experiences.
- Simulation of carbon dioxide emission levels when no data is available: This functionality will allow for the simulation of carbon dioxide emission levels based on available data and artificial intelligence algorithms, even in the absence of specific data. This will enable companies to better monitor and manage their carbon footprint, which is increasingly important in the context of sustainable development and environmental stewardship.
- Automated creation of ESG scorecards for suppliers: This functionality will enable the automatic creation of ESG (Environmental, Social, Governance) scorecards for suppliers based on data regarding their sustainability-related activities. This will allow companies to assess and monitor their suppliers’ compliance with sustainable development guidelines and corporate social responsibility more quickly and effectively.
- Intelligent summary of typical supplier errors: This functionality will enable the identification and summarization of typical errors made by suppliers in the purchasing process, allowing companies to respond more quickly and eliminate problems. This will make purchasing processes more efficient and error-free, potentially saving time and resources.
Truly intelligent summary
In the light of the above facts, it seems that the future of purchasing lies in the applications of artificial intelligence, which will not only increase the efficiency of decision-making but also streamline collaboration between entities in the supply chain. For companies seeking a competitive advantage, investing in AI in purchasing is becoming increasingly essential.
However, to harness the full potential of artificial intelligence, it is necessary to have a solid digital foundation in the form of a digitized Source to Pay process. Without this foundation, AI may only be an ineffective tool that fails to deliver expected benefits and does not support effective management of corporate expenditures.
Therefore, companies should focus not only on implementing new technologies but also on modernizing existing purchasing processes to ensure optimal conditions for leveraging the full potential of artificial intelligence.
In summary, artificial intelligence has the potential to revolutionize the way purchasing is managed, enabling companies to achieve higher levels of efficiency, cost optimization, and better control over business operations. However, to harness the full potential of artificial intelligence, it is necessary to have a solid digital foundation and understanding of the different types of AI and their practical applications in purchasing processes.
