The growing availability of domain-specific industry reports presents an opportunity for enhanced knowledge acquisition and management. This paper examines the application of generative AI in extracting structured semantic knowledge from business texts, utilizing the 2025 Logility Market Report as a case study. By applying a large language model (LLM) and prompt engineering, we automatically generated RDF triples representing key entities and relationships in AI-driven supply chain management. The methodology included both text-based extraction and OCR-based processing of an infographic, demonstrating the potential of multimodal LLMs for transforming visual data into machine-readable structures. The extracted triples were evaluated in terms of semantic coherence, completeness, and interpretability, and compared against a rule-based NLP pipeline implemented in spaCy. The results show that GPT-4 produced richer and more contextually accurate triples, while traditional methods struggled to capture implicit relations and numerical context. Although the graphs lacked formal ontological grounding, they provide a foundation for knowledge management applications and highlight the importance of hybrid approaches that combine LLM-based extraction with ontology alignment. The study also discusses methodological trade-offs between browser-based and API-based access, and outlines limitations such as reproducibility, formatting sensitivity, and scalability. Overall, the findings support the role of generative AI in augmenting semantic knowledge management in business environments.