Explore LLM architectures, general-purpose limitations, and why adapting models through Fine-Tuning and RAG is the real engineering challenge.