Exploring the Potential of Multi-Agent AI Systems in Ecosystemic Data Exchange for EUDR Compliance
Werahara Arachchige, Nilushi (2026)
2026
Master's Programme in Computing Sciences and Electrical Engineering
Informaatioteknologian ja viestinnän tiedekunta - Faculty of Information Technology and Communication Sciences
Hyväksymispäivämäärä
2026-06-10
Julkaisun pysyvä osoite on
https://urn.fi/URN:NBN:fi:tuni-202606107205
https://urn.fi/URN:NBN:fi:tuni-202606107205
Tiivistelmä
The European Union Deforestation Regulation (EUDR) introduces due diligence requirements under which companies must ensure that commodities entering the EU market are not linked to deforestation. This creates a need for ecosystem-wide data exchange across multiple stakeholders within a supply chain. However, current data management practices are characterized by fragmented information systems across organizations, incompatible data formats, and limited traceability of products to their origin. In addition, many supply chains often focus on operational data, while broader ecosystemic data required for compliance remains incomplete.
This thesis addresses these challenges by proposing a multi-agent system (MAS) architecture for coordinated ecosystemic data exchange. The approach represents each supply chain stage as an autonomous AI agent responsible for managing data retrieval, validation, and traceability processes. The multi-agent system coordinates agent interactions, while a unified data schema is introduced to enable standardized communication between agents. In addition, deterministic functions are incorporated within agents to perform core tasks such as data validation and transformation, enabling reliable and consistent execution of the workflow.
A functional prototype was developed in local environment and evaluated using a simulated supply chain dataset to assess traceability, interoperability, and system performance across different evaluation scenarios. The results indicate that the proposed MAS-based architecture supports consistent data exchange and enhances end-to-end traceability in a controlled evaluation setting. The findings suggest that MAS provides a scalable approach for addressing EUDR-driven data exchange challenges.
This thesis addresses these challenges by proposing a multi-agent system (MAS) architecture for coordinated ecosystemic data exchange. The approach represents each supply chain stage as an autonomous AI agent responsible for managing data retrieval, validation, and traceability processes. The multi-agent system coordinates agent interactions, while a unified data schema is introduced to enable standardized communication between agents. In addition, deterministic functions are incorporated within agents to perform core tasks such as data validation and transformation, enabling reliable and consistent execution of the workflow.
A functional prototype was developed in local environment and evaluated using a simulated supply chain dataset to assess traceability, interoperability, and system performance across different evaluation scenarios. The results indicate that the proposed MAS-based architecture supports consistent data exchange and enhances end-to-end traceability in a controlled evaluation setting. The findings suggest that MAS provides a scalable approach for addressing EUDR-driven data exchange challenges.