A large language model — the neural network behind tools like Claude and ChatGPT, trained on huge amounts of text to predict what comes next.
At its core, an LLM does one thing: given some text, predict the next token (word fragment). Trained on enough text, that single skill turns out to cover writing, coding, translation, and reasoning — because predicting what comes next in a math proof or a Python file requires actually modeling how those things work.
Everything else on this site — agents, RAG, fine-tuning, evals — is machinery built on top of that next-token engine.