Text to SQL is a burgeoning field that focuses on converting natural language questions into SQL queries, providing a more intuitive interface for interacting with complex database schemas. This area has been significantly influenced by the rise of Large Language Models (LLMs) such as OpenAI's ChatGPT and Llama, which have become the standard approach for tackling these conversions. Despite advancements, these models face challenges, including hallucination when confronted with unfamiliar databases and difficulties with complex SQL syntax. Various datasets, like Spider and WikiSQL, are used for training and evaluating these models, with synthetic datasets also gaining traction. Evaluation metrics for these models include content-matching and execution-based evaluations, with execution-based methods providing a more reliable assessment. Models use techniques such as in-context learning and fine-tuning, with data augmentation and decomposition methods improving performance. Agentic frameworks, like MAC-SQL, address limitations of LLMs by incorporating auxiliary agents to refine SQL generation. The field continues to evolve with ongoing research into optimizing smaller language models and enhancing pipeline robustness through agentic approaches.