![IBacademy_Logo_blau[496]](https://www.ibsolution.com/hs-fs/hubfs/IBacademy_Logo_blau%5B496%5D.jpg?width=200&name=IBacademy_Logo_blau%5B496%5D.jpg "IBacademy_Logo_blau[496]"){kind=logo}
High data quality is becoming increasingly important for business success. However, manual maintenance of master data is time-consuming and prone to errors. Companies are therefore looking for ways to make their master data management more efficient and reliable with the help of innovative technologies. This inevitably brings artificial intelligence (AI) into focus. Integrating AI-based processes into master data maintenance enables a significant leap in efficiency and improves data quality. The key questions here are: Where is the best starting point for companies? How can AI be used effectively?
SAP MDG: the right decision for your master data
Use case for automatic data extraction
Basically, there are various use cases for master data management in which AI can be a helpful tool. An initial solution developed by IBsolution focuses on the use of artificial intelligence for the intelligent analysis and extraction of data from unstructured documents. Various AI services are used for this purpose:
-
Optical character recognition (OCR) for automated data extraction
-
Natural language processing (NLP) for intelligent data processing
-
Large language models (LLMs) for data optimization
In this example, SAP Master Data Governance (MDG) acts as the central platform for master data management. The advantage here is that existing processes mapped in workflows are retained. The AI services required for automatic data recognition and interpretation are connected via standardized interfaces, meaning that users continue to experience the familiar look and feel of SAP MDG. The REST API ensures secure and scalable integration into the existing system landscape.
What to consider with regard to AI models
In order for artificial intelligence to deliver the greatest possible added value, it needs clear inputs. This means that the more structured and precise the input, the better the results. In this respect, prompt engineering is of crucial importance. When companies draw on the expertise of external specialists such as IBsolution, they benefit from optimally defined and formulated inputs for the AI system, which lead to the best possible results.
Another characteristic of this use case is that security and data protection can be flexibly controlled. Both the use of cloud services and access to local instances are possible.
The scenario described uses the ChatGPT AI model. In principle, however, AI models from other providers can also be used for this scenario. Billing is based on tokens. The AI models operate with “word units” that are divided into tokens. The computing time and price depend on these tokens. The tokens are composed differently depending on the language.
Scenario: Create or change business partners in SAP MDG
In our example, AI is used when creating or changing a business partner (customer, supplier) in SAP MDG. In the first scenario (creation), the business partner's address is to be automatically transferred from a document, such as an invoice or letterhead, to the further creation process in order to save time and eliminate errors. In the second scenario (change), a new address from a document is to be loaded into the existing business partner data record.
The AI-supported process is the same in both cases: The user starts a change request as a workflow in SAP MDG, whereupon a screen appears for entering the basic data. In the next step, the user uploads the document and then calls up the corresponding AI service (OCR, NLP, LLM) in the background. After the information has been automatically extracted and processed, the user can decide which data should actually be transferred to the corresponding fields. Following this validation, the user can continue working in SAP MDG as usual. Thanks to AI, the processing time for creating or changing business partners is reduced from several minutes to a few seconds. At the same time, the error rate is significantly reduced compared to manual entry. This makes it possible to establish standardized and validated data quality.
Conclusion: An important lever for automated master data maintenance
The use of artificial intelligence makes master data processes more efficient, reliable, and future-proof. The automation of repetitive tasks results in increased efficiency and higher productivity. AI-supported verification mechanisms lead to more consistent and reliable data, while central interfaces and standardized processes ensure high scalability.
However, for artificial intelligence to actually generate added value for master data management, a number of prerequisites must be met. On the one hand, a good basis of training data for the AI models is required; on the other hand, the results of the AI must be regularly checked and improved through monitoring. Furthermore, roles and processes must be clearly defined so that it is easy to understand who is responsible for which tasks. Artificial intelligence is not a sure-fire success, but under the right conditions, it is an important lever for optimizing master data management.
