ChIRAAG is a novel framework designed to generate System Verilog Assertions (SVAs) from natural language specifications using Large Language Models (LLMs), specifically OpenAI GPT4. The framework aims to streamline the Assertion Based Verification (ABV) process by reducing the time and effort traditionally required for expert-driven SVA formulation. ChIRAAG involves two main stages: formatting design specifications into a standardized JSON format and using LLMs to generate raw assertions. These raw assertions are then validated through simulation, and any errors are refined based on the simulation logs. The framework has been evaluated on OpenTitan designs, showing that LLMs can generate correct assertions with minimal manual intervention. The study also highlights the potential of ChIRAAG in detecting bugs in the implementation and generating more comprehensive assertions compared to traditional methods. The framework's effectiveness is demonstrated through experiments, where only 27% of LLM-generated raw assertions required refinement, and it can generate assertions for each design in less than 15 seconds. Future work will focus on improving the consistency and completeness of the generated assertions.ChIRAAG is a novel framework designed to generate System Verilog Assertions (SVAs) from natural language specifications using Large Language Models (LLMs), specifically OpenAI GPT4. The framework aims to streamline the Assertion Based Verification (ABV) process by reducing the time and effort traditionally required for expert-driven SVA formulation. ChIRAAG involves two main stages: formatting design specifications into a standardized JSON format and using LLMs to generate raw assertions. These raw assertions are then validated through simulation, and any errors are refined based on the simulation logs. The framework has been evaluated on OpenTitan designs, showing that LLMs can generate correct assertions with minimal manual intervention. The study also highlights the potential of ChIRAAG in detecting bugs in the implementation and generating more comprehensive assertions compared to traditional methods. The framework's effectiveness is demonstrated through experiments, where only 27% of LLM-generated raw assertions required refinement, and it can generate assertions for each design in less than 15 seconds. Future work will focus on improving the consistency and completeness of the generated assertions.