Hamra, Mumen (2026) Development and integration of AI-supported agents with databases for the automation of business processes. Masters thesis, Universität Rostock.
Full text not available from this repository.Abstract
This thesis presents the design and implementation of an AI-supported agents architectureintegrated with an enterprise database to automate business processes within an organizational environment. Public, general-purpose AI services are often unsuitable for handling sensitive business information due to concerns regarding data privacy, regulatory compliance, and limited domain-specific reliability. Motivated by these challenges and by the growing adoption of Large Language Models (LLMs) capable of understanding and generating natural language, this work investigates how enterprise-specific AI agents can interact with internal data sources and workflow systems. The conceptual contribution of this thesis lies in the development of an architecture that enables an AI agent to interpret user requests, access structured enterprise data, support decision-making, and initiate workflow-based actions in a controlled and secure manner. To manage process complexity, the architecture adopts a multi-agent design in which specialized subordinate agents execute well-defined sub-tasks and return their results to the primary agent, which retains responsibility for the overall workflow orchestration. The approach integrates natural language interaction with rule-based process execution, creating a unified framework through which employees can obtain relevant information and trigger operational tasks using conversational input. To demonstrate the feasibility of the concept, the proposed solution implements a system architecture in which Microsoft Copilot Studio–based agents operate entirely within the organization’s-controlled infrastructure and interact with an Oracle database, BPMN-modeled processes and workflows automation. Communication with external systems is facilitated through REST APIs and the MCP protocol, secured via OAuth 2.0-based authorization. The results demonstrate that tasks traditionally requiring extensive manual effort can be automated in a reliable, secure, and scalable manner. The solution significantly enhances process efficiency, data accuracy, and operational consistency. This thesis therefore provides a practical and extensible foundation for organizations seeking to integrate AI-driven agents into existing IT systems while ensuring adaptability, security, and compliance.
| Item Type: | Thesis (Masters) |
|---|---|
| Subjects: | Autorenart > Studentische Arbeiten > Masterarbeit Autorenart > Studentische Arbeiten |
| Depositing User: | Dbis Admin |
| Date Deposited: | 06 Oct 2026 11:24 |
| Last Modified: | 06 Oct 2026 11:24 |
| URI: | https://eprints.dbis.informatik.uni-rostock.de/id/eprint/1160 |
Actions (login required)
 |
View Item |
