Abstract:

In dynamic marketing environments, the ability to adapt campaigns in real time to changing external conditions and consumer interests is increasingly critical. This paper presents an LLM-based multiagent system designed to support real-time marketing adaptation by integrating weather data and online search trends. The system consists of specialized agents responsible for data collection, integration, data analysis and campaign analysis. By combining external real-time data with company-specific information, the architecture enables automatic identification of relevant correlations and generation of marketing recommendations tailored to current conditions. Use cases illustrate how the system enhances decision-making speed and personalization by providing useful information and content tailored to local trends and weather. This thesis contributes to the field by introducing a multi-agent system architecture that operationalizes large language models for adaptive data-driven marketing.