Zero-shot object detection (ZSD) is a technique that allows for the identification of target object classes without requiring labeled training data, an approach that can save time and money by reducing the need for expensive human labeling. This method utilizes OpenAI's CLIP embeddings to match the visual embeddings of regions of interest in an image with text descriptions of the target classes. A dynamic threshold determines the presence of an object based on the distance between visual and text embeddings, and the choice of text prompts can significantly impact accuracy, with the option to average multiple text embeddings for improved results. The effectiveness of ZSD is demonstrated through tasks such as detecting "jumping cats" versus "sitting cats," classifying real estate images as houses or apartments, and identifying discount banners, showcasing its potential in scenarios where labeled data is costly to obtain. The results, including a low false positive rate, highlight the capabilities of ZSD using CLIP, and ongoing development aims to enhance this technology further.