The rapid adoption of large language models (LLMs) in knowledge-intensive professions has outpaced the development of rigorous methods for assessing their performance on domain-specific tasks. While general-purpose benchmarks abound, the project management and business analysis disciplines lack a structured, reproducible means of measuring how well such models support core professional processes. This thesis addresses that gap by designing and operationalising a methodological framework for the automated benchmarking of LLMs across project management and business analysis activities. The framework is grounded in a practitioner survey that identified and validated seven macro-processes central to the two disciplines. From these processes, twelve operational tasks were derived, spanning estimation, scheduling, dependency identification, risk and stakeholder analysis, requirements elicitation, and related activities.