Manual Convolutional Neural Network (CNN) design for image classification tasks is costly, complex, and time-consuming. Evolutionary Neural Architecture Search (ENAS) provides an automated alternative in searching for effective CNNs, with Genetic Algorithms (GAs) offering flexible search strategies to explore vast, complex search spaces under well-defined encodings. This proposal investigates the use of Large Language Models (LLMs) to augment a GA-based ENASframework to improve search performance and computational efficiency. This research examines five specific areas: the use of LLMs as classifier and repair mechanisms within the GA in a ResNet-like search space, the generalisability of the framework to additional CNN architectures, the introduction of LLM-based dataset-awareness into the search process, the use of constrained in-algorithm prompt editing via LLMs to improve search performance, and the application of agentic augmentation to improve speed and overall effectiveness. The expected contribution is a more adaptive and efficient GA-based evolutionary framework for CNNarchitecture search under practical compute constraints.