APPLICATION OF AI SOLUTIONS FOR THE DEVELOPMENT OF BOTS FOR COLLECTING AND STRUCTURING BUSINESS INFORMATION IN THE SAP ENVIRONMENT

ABSTRACT

The article explores approaches to applying artificial intelligence solutions for the development of intelligent bots integrated into the SAP environment. It analyzes the architectural, technological, and functional aspects of such systems, including the use of cloud platforms, API integrations, and natural language processing algorithms. It is emphasized that AI-based bots are capable of efficiently collecting, interpreting, and structuring unstructured and semi-structured business information in real time. For empirical validation of the proposed theoretical assumptions, the study presents an experiment comparing different approaches to processing reference data within SAP. It is concluded that the implementation of intelligent agents in corporate IT systems contributes to increased productivity, cost reduction, and improved quality of managerial decision-making.

АННОТАЦИЯ

В статье рассматриваются подходы к применению решений на базе искусственного интеллекта для разработки интеллектуальных ботов, интегрированных в САП-среду. Анализируются архитектурные, технологические и функциональные аспекты таких систем, включая использование облачных платформ, АПИ-интеграции и алгоритмов обработки естественного языка. Подчеркивается, что ИИ-боты способны эффективно собирать, интерпретировать и структурировать неструктурированную и полуструктурированную бизнес-информацию в режиме реального времени. Для эмпирической проверки выдвинутых теоретических положений в работе представлен эксперимент по сравнению различных подходов к обработке справочных данных в САП. Делается вывод, что внедрение интеллектуальных агентов в корпоративные ИТ-системы способствует росту производительности, снижению издержек и улучшению качества управленческих решений.

Keywords: SAP, artificial intelligence, artificial intelligence bots, data processing, cloud technologies, automation, business analytics.

Ключевые слова: САП, искусственный интеллект, ИИ-боты, обработка данных, облачные технологии, автоматизация, бизнес-аналитика.

Introduction

Computerization of business management systems necessitates automated collection, processing, and formatting of huge volumes of business data in real time. One of the widely used platforms in large corporations for complementing operational, financial, and logistics functions is SAP. Nevertheless, traditional approaches in managing data within the SAP platform tend to entail manual handling, which slows down decision-making and enhances the chances of errors. In this context, the need to harness artificial intelligence (AI) solutions, in particular, smart bots that can automatically extract, comprehend, and arrange data from sources such as SAP and external cloud environments, is growing.

The purpose of this study is to analyze the potential of AI technologies for developing bots integrated into the SAP ecosystem, enabling intelligent data processing through methods of natural language processing (NLP), parsing, and logical data structuring. This includes examining the architectural and functional integration of such systems with cloud-based AI services, SAP API, and modular processing pipelines. To validate the feasibility and practical effectiveness of such solutions, the study puts forward a hypothesis that integrating AI-based models into the architecture of smart bots significantly improves the speed, accuracy, and reliability of business data processing compared to traditional or RPA-only methods. The research additionally aims to demonstrate that these improvements contribute to higher decision-making quality and reduced operational costs, particularly in large-scale enterprise environments. This hypothesis is tested and confirmed through an applied experiment presented in the corresponding section of the article.

Methods

The study comprises theoretical and practical components, each employing a corresponding set of scientific and applied methods.

In the theoretical part, a comparative analysis was conducted of the architectural and functional characteristics of intelligent bots integrated into the SAP environment through comparative analysis methods, induction, deduction, and systematization of scientific literature. Contemporary concepts of the use of cloud computing, API integrations, and natural language processing algorithms in corporate information systems were investigated. Specific focus was placed on comparing conventional automation (RPA) and cognitive AI bots in terms of flexibility, scalability, and the ability to handle unstructured data. Based on this comparison, a hypothesis was formulated regarding the higher efficiency of AI-based solutions.

The empirical part of the study included experimental confirmation of the hypothesis developed. Applied systems modeling, observation, quantitative analysis, and experimental comparison of the results of three data processing methods – manual input, template RPA, and the developed AI bot – were used at this stage. The performance of each method was measured against metrics such as average document processing time, percentage of successfully normalized data, and errors per 1,000 documents. Reproducibility was assured by threefold repetition of each experiment, and metric collection and visualization were achieved by monitoring systems such as Grafana and Prometheus.

Results

The outcomes of the study present two directions of the research – conceptual-analytical and experimental-applied. The former is characteristic of generalized conclusions made through analysis of the opportunities for AI integration in SAP system design. The latter presents results of practical implementation and testing. The organization permits a complete demonstration of the theoretical justification of the presented approach and its applied correctness in real business processes.

Technological foundations for developing AI bots in the SAP environment

To supply terminological specificity and accurate understanding of the functions of smart components in the architecture at hand, a need arises to describe what AI bots are and how, in fact, they are distinct from traditional automated solutions. AI bots are computer programs that use AI algorithms to perform tasks on information extraction, analysis, and interpretation without the need for human intervention.

Unlike traditional RPA bots, which are strictly tied to pre-programmed rules and circumstances, AI bots possess cognitive capabilities – they can understand natural language, classify data, identify patterns, and react to changing situations. AI bots can be utilized as standalone modules or integrated in the IT core of corporate systems (e.g., SAP), making it possible to process business data in real time with intelligent capabilities. Their use is particularly relevant for operations involving processing of unstructured data, human–machine interaction, and support for managerial decision-making. In 2024, the global AI chatbot market was valued at $8,6 billion, according to The Business Research Company (fig. 1).

Figure 1. AI chatbot market size, billion dollars [1]

The effective use of AI bots in a corporate information environment is impossible without a deep understanding of the architectural features of the target platform, in this case – SAP. It is a modular ERP environment designed for centralized real-time management of business processes [2]. Its architecture includes both on-premise solutions and cloud-based components (such as SAP S/4HANA Cloud and SAP Business Technology Platform), enabling flexible integration of external intelligent modules (fig. 2).

Figure 2. SAP system architecture

The SAP API infrastructure, including RESTful, OData, SOAP, and RFC protocols, enables external system communication with full read/write access across SAP modules. The SAP API Hub serves as a central catalog for standardized data and service access, supporting seamless integration of AI models into ERP workflows.

Cloud platforms like Azure, GCP, and AWS provide the environment for deploying AI models using pre-trained algorithms, AutoML, and data pipelines. Integration is achieved via tools such as SAP Cloud Connector, allowing data to flow from SAP to the cloud for processing and back in structured form.

Together, open SAP API and scalable cloud infrastructure form the core technological basis for intelligent, flexible, and high-performance automation with AI bots.

AI components of intelligent bots: functional structure and data processing methods

The development of smart agents with the capability to independently gather and process business data involves the use of diverse algorithms and architecture solutions in AI. In systems with SAP integration, AI components perform cognitive functions, which opens up possibilities for shifting from structured field processing towards semi-structured and unstructured data processing from internal or external sources. These components serve as functional modules, each of which deals with a specific stage of information transformation – from raw data acquisition to logical arrangement and subsequent passage to business applications (table 1).

Table 1.

AI components of intelligent bots in the SAP environment [3]

The modular architecture of AI bots ensures high adaptability, allowing each component to be independently scaled or customized for specific organizational needs. Transformer-based NLP models enable precise understanding of inputs and documents, while rule-based parsers reduce errors in processing semi-structured data. This combination supports closed-loop, intelligent interaction between users, SAP systems, and analytics platforms, enhancing data accuracy, processing speed, and the scope of automation through context-aware and adaptive operations.

Development stages of AI bots for intelligent business information processing in the SAP environment

The development of AI bots integrated into corporate IT systems (particularly SAP) is a multi-stage engineering and analytical process that combines machine learning methods, software architecture design, data handling, and interface technologies (table 2).

Table 2.

 Stages of AI bot development for integration into the SAP environment [4, 5]

Thus, the development of AI robots is a complex, multi-step procedure based on AI methods, enterprise engineering, and integration of systems. Effective deployment relies upon an interdisciplinary approach, high formalization of requirements, and strict adherence to rules of reliability and interpretability of solutions.

Advantages of implementing AI bots in the SAP environment

The incorporation of intelligent AI-based bots within the corporate SAP environment has a significant impact on business information processing efficiency and accuracy. Compared to data collection and analysis via conventional methods relying on human effort or scripted, strictly defined scripts, AI-based bots bring cognitive adaptability, self-improvement, and adaptability towards unstructured data. This comes down to business analytics operations being faster, less prone to errors, with improved quality of decisions, and easier manageability of the overall process (table 3).

Table 3.

Comparison of traditional and AI-based approaches to business information processing in SAP [6,7]

The use of AI bots in business operations enhances efficiency, reduces costs, and improves client engagement. McKinsey, for instance, developed an internal generative AI tool, Lilli, to access its extensive knowledge base. Over 70% of employees use it for information retrieval and decision-making, cutting operational time by 30% and boosting productivity [8]. Intel applied AI bots with SAP S/4HANA to automate data processing in finance and operations, enabling real-time data integration from diverse sources, which improved reporting speed and decision accuracy. Similarly, Accenture implemented AI bots in procurement using SAP Ariba and Fieldglass, achieving a 77% rise in supplier onboarding and a 7% increase in automated invoice processing, with overall gains in efficiency [9].

Evaluation of the effectiveness of intelligent bots for the extraction and normalization of business reference data in the SAP environment

To confirm the hypothesis that the use of AI models in the architecture of intelligent bots significantly improves the efficiency of business data processing in the SAP environment, a practical experiment was conducted. The objective was to quantitatively assess the increase in processing speed, completeness, and reliability compared to manual methods and classical robotic automation.

As a case study, the task involved extracting, normalizing, and integrating supplier and product reference information received through email attachments and commercial proposals in various formats (PDF, DOCX, XLSX). Such data typically cannot be entered directly into SAP without prior processing – they require text recognition, key attribute extraction (e.g., name, material number, price, currency, contact data), and mapping to standard SAP data structures (e.g., LFA1, MARA, T001W).

The following architecture was used to implement the intelligent bot:

• Bot platform: a custom Python-based bot deployed on the SAP Business Technology Platform and connected to SAP S/4HANA via OData and RFC interfaces.
• OCR module: Tesseract library for extracting text from attachments.
• NLP: a fine-tuned BERT model (bert-base-cased) adapted to the terminology of commercial offers and product descriptions.
• Analysis and normalization: Python scripts using Pandas and spaCy for entity recognition and field mapping to SAP structures.
• Integration: SAP Integration Suite with IDoc and REST communication.
• Monitoring: Grafana + Prometheus for tracking performance and logging system metrics.

Results

To assess the effectiveness of the proposed AI-bot framework in real-world applications, a comparative trial was conducted using a set of 1,000 business documents with supply and product reference information. The trial compared and analyzed three methods for processing information in the SAP system: manual entry by experts, RPA automation via templates, and the smart AI-based bot developed in this study. Each method was tested in triplicate, and means of three performance metrics: document processing time, rate of successful data normalization, and error/1,000 records, were measured.

The experimental dataset consisted of 1,000 commercial documents collected over a one-month period from the procurement department’s email channel. Three data processing approaches were evaluated:

• Manual data entry by SAP specialists;
• Classic RPA bot using template-based extraction;
• Intelligent AI-based bot as described above.

Each test series was repeated three times, and average values were calculated for the following metrics: average processing time per document (in seconds), percentage of successfully normalized records, and number of errors per 1,000 documents (including duplicate entries, field omissions, and formatting violations). The obtained data are presented in the table 4.

Table 4.

Comparative analysis of business reference data processing methods in SAP

Analysis of the results in the table demonstrates that the intelligent AI-based bot shows a clear advantage across all evaluated metrics. The average processing time per document is reduced by almost sixfold compared to manual input, while successful normalization increases to 96,9%, and the number of errors decreases by more than five times. These results confirm the superior numerical performance of the AI-based approach across all evaluated metrics.

Discussion

The experiment confirms the validity and value of using AI solutions to enable auto-gathering and normalization of business reference information with smooth integration into SAP systems. The solution facilitates enhanced productivity, accuracy, and scalability of data processing operations in enterprise IT environments.

The secret to the AI bot's success is in its modularity and transformer-based language models, which can understand business context and perform robust entity recognition even when input data are noisy and inconsistent. Unlike classical RPA systems based on rigid templates and pre-coded logic, AI models offer flexibility and responsiveness to actual document variations, reducing the need for manual template updates and rule maintenance.

From a commercial perspective, the deployment of such intelligent agents leads to lowered operational costs, especially in terms of improved data quality and faster processing time, thereby enabling faster and better-informed decision-making. These benefits are most essential for procurement and supply chain operations, where a lag or error in the reference data will lead to the stoppage of the operation. Moreover, the ability to leverage the AI-bot architecture across different SAP modules – such as finance, logistics, or HR – opens up large potential for enterprise-scale optimization.

Although effective, AI models must be retrained periodically to maintain ongoing accuracy for changing business situations. Finally, the quality of document preprocessing – particularly OCR performance – is frequently decisive in determining success of downstream data extraction.

Additional work may be in the lines of introducing multilingual content processing, adaptive learning processes, and integrating more closely with newer SAP services such as predictive analysis and machine learning orchestration within the SAP BTP domain.

Conclusion

Adoption of AI technologies for designing smart bots integrated into the SAP environment is one of the promising means in digital transformation of business processes. The bots facilitate automated collection, semantic understanding, and structuring of information from various sources, including unstructured and semi-structured information. With the use of advanced NLP, machine learning, and cloud computing powers, accuracy in data interpretation is optimized, processing is reduced, and analytics reliability and transparency are improved. The technology environment – integrating SAP environments, AI building blocks, and cloud infrastructure – makes scalable, elastic, and fault-resilient solutions available to deploy that can operate in real time. The results of the conducted experiment confirm the practical effectiveness of AI-based bots in accelerating information processing and improving the quality of data integration in SAP systems.

Real-world case studies demonstrate that integrating AI bots into corporate IT infrastructure yields substantial dividends: increased cost savings, accelerating business work, and general productivity gains. Most significantly, these systems possess a retraining and learning ability that allows them to be immune to changing business environments and input data formats. Therefore, intelligent bots become more than automation technology in themselves but rather as key elements of a cognitive IT environment that facilitates strategic business expansion and digital maturity.

References:

1. AI chatbot Global market report 2025 / The Business Research Company // URL: https://www.thebusinessresearchcompany.com/report/artificial-intelligence-ai-chatbot-global-market-report (date of access: 26.05.2025)
2. Gunturu N.S.R. An Overview on SAP S/4HANA Deployment Options and Transition Paths // International Journal of Advanced Research in Science, Communication and Technology. – 2024. – P. 209-216. DOI: 10.48175/IJARSCT-18826
3. Shekhar S. Investigating the integration of Artificial Intelligence in enhancing efficiency of distributed order management systems within SAP environments // Applied Research in Artificial Intelligence and Cloud Computing. – 2024. – № 7 (5). – P. 11-27.
4. Jamithireddy N.S. Automation of SAP ERP Processes Using Agentic Bots and UiPath Framework // Research Briefs on Information and Communication Technology Evolution. – 2025. – № 11. – P. 62-81. DOI: 10.69978/rebicte.v11i.212
5. Garifullin R. Development and implementation of high-speed frontend architectures for complex enterprise systems // Cold Science. – 2024. – № 12. – P. 56-63.
6. Borodin I. The impact of integrated management information systems on cost control efficiency in large construction and engineering projects // International Journal of Latest Engineering and Management Research. – 2025. – № 10 (3). – P. 24-27.
7. Blazhkovskii A. Collecting metrics for continuous platform monitoring // Universum: technical sciences: electronic scientific journal. – 2025. – №3 (132). – P. 10-15.
8. Meet Lilli, our generative AI tool that’s a researcher, a time saver, and an inspiration / McKinsey // URL: https://www.mckinsey.com/about-us/new-at-mckinsey-blog/meet-lilli-our-generative-ai-tool (date of access: 26.05.2025)
9. SAP Innovation Awards 2024 – Accenture / SAP // URL: https://www.sap.com/sweden/documents/2024/01/96128ca1-a57e-0010-bca6-c68f7e60039b.html (date of access: 26.05.2025)