Skip to content
Zarif Automates

Large Language Model

A large language model (LLM) is an AI trained on huge amounts of text to predict the next word, which lets it read, write and answer questions.

A large language model, or LLM, is a program trained on a huge pile of text until it got very good at guessing what word comes next.

Example

Paste in a meeting transcript and ask for a three-bullet summary. The model writes the bullets one word at a time. Each guess is shaped by the transcript, your request, and everything it has written so far. (Strictly, it works in tokens, which are pieces of words, but "words" is close enough.)

That one skill, predicting the next word well, turns out to cover reading documents, drafting emails, summarizing meetings, sorting data and writing content.

Well-known LLMs include OpenAI's GPT models, Anthropic's Claude, Google's Gemini and Meta's Llama.

How it works

During training, the model reads billions of text samples from books, websites, code and other sources, and practices predicting the next word. Along the way it picks up grammar, facts and patterns of reasoning.

What it learns is stored in billions of numbers called parameters. Training nudges each one up or down until the guesses get better. Most LLMs are built on a design called the transformer, which lets the model weigh every earlier word when it picks the next one.

The answer you get depends on the model, what it was trained on, and how clearly you asked. Three common ways to get better results for a specific job:

  • Prompt engineering: writing clearer instructions.
  • Retrieval-augmented generation (RAG): looking up relevant documents and handing them to the model along with the question.
  • Fine-tuning: training the model a bit more on your own examples.

In automation

In a workflow, an LLM is usually one step among several. An n8n workflow might send each new support ticket to an LLM, which labels how urgent it is, drafts a reply, and picks the team it should go to. The rest of the workflow then does the routing.

  • AI Automation: using AI, including LLMs, to handle steps in a workflow
  • Prompt Engineering: writing instructions that get the output you need
  • Agentic AI: AI that works toward a goal in steps, built on LLMs
  • n8n: an automation tool with built-in LLM nodes