Unit testing, a cornerstone of software development, is undergoing a transformation thanks to the power of Large Language Models (LLMs). Imagine AI effortlessly crafting those crucial assert statements that validate code functionality. This is the promise of "Chat-like Asserts Prediction" (CLAP), a groundbreaking approach leveraging the prowess of LLMs. CLAP doesn't just generate asserts; it engages in a dynamic conversation with the code, refining its predictions through feedback from the Python interpreter. Think of it as an AI pair programmer specializing in unit tests. This innovative technique uses a "Chain-of-Thought" prompting, guiding the LLM through the reasoning process of crafting effective assertions. The results are impressive. CLAP achieves remarkable accuracy in generating both single and multiple assert statements, outperforming existing methods by a significant margin. Moreover, CLAP's impact goes beyond mere automation; its generated assertions often improve code readability. The team's submitted code changes, based on CLAP's predictions, have been embraced by real open-source projects, a testament to its practical value. While CLAP excels in generating single assert statements, multiple assert generation presents unique challenges due to code length and complexity. This suggests a promising area for future research. CLAP also showcases its adaptability to diverse LLMs, demonstrating its robust design. However, the current metric for evaluating "meaningful" assertions presents a potential limitation. Despite this, the study highlights the revolutionary potential of LLMs in automating unit testing and improving code quality. This technology promises to be a game-changer for developers, freeing them from the tedious task of writing unit tests and empowering them to focus on building exceptional software.