The text discusses the evolution of table extraction techniques, from traditional methods to the use of Large Language Models (LLMs). It highlights the limitations of traditional approaches and the potential of LLMs in handling complex table formats. The article introduces key LLMs, such as GPT-4o, Gemini, and Mistral-Nemo-Instruct, and demonstrates their capabilities in extracting tables from documents using OCR and prompt engineering. The text also explores the challenges associated with LLM-based extraction, including repeatability, black box nature, hallucination, scalability, cost, privacy, and the need for fine-tuning. Nanonets' approach to table extraction is discussed, which involves converting OCR output into a rich text format, using pre-trained models, and providing a user-friendly interface. The article concludes that LLMs offer flexible capabilities in understanding context but are not as consistent as traditional OCR methods, and tools like Nanonets are pushing the boundaries of what's possible in automated table extraction.