This paper focuses on the design, construction and application of an AI agent, the Moodle survey tool, for the purpose of evaluating higher degree online courses, and which, it is proposed, will take the form of a module built into Moodle courses. The project is informed by the critical realist philosophy, which focuses on practical interventions in social life. The contention of this paper is that, if the deep-level structure of the mechanisms which govern social systems such as education can be identified, the systemic models thus formed can be used to gain insight into curriculum design and evaluation. The course evaluation module is therefore based on a systemic model of course design which identifies the functions needed for teaching and learning to occur; it does not prescribe what effective teaching and learning might constitute, but identifies the systemic functions needing to be carried out in pedagogical processes. The Moodle survey tool was set up according to the framework given in the model, so that it might elicit, collate and analyse participants’ responses to online courses. Intended participants will ultimately include not only students, but also lecturers and course designers. The project provides an exemplar of digital/human collaboration which might form the basis of later projects using AI agents in more complex scenarios. The contribution to knowledge is thought to lie in demonstrating the practical application of curriculum theory to provide feedback that not only can effect mass evaluation of online courses but can also be used to suggest refinements to specific courses.