Student will explore current state of the art in the domain of using Large Language Models (LLM) for automatic text generation. Current LLMs have relatively small context window so usage of various approaches will be required to obtain a longer coherent text. Student will apply processes from software development in order to test if they are applicable also when programming using natural language. The goal of this thesis is to create bachelor thesis solely from title and problem definition, so prompt sets have to be generic enough. For the "generated" thesis student will obtain reviews and he will either fix the issues or explain limitations of selected approach. The obtained knowledge will be used for creating prompts that will generate bachelor thesis. As we expect that generated text will contain various flaws, student will describe them and propose solutions if possible.