Definition
An AI model trained on huge text corpora, able to understand and generate natural language, the building block of AI assistants.
An LLM (Large Language Model) is a neural network trained to predict text, which in practice can write, summarize, translate, analyze documents and reason. GPT (OpenAI), Claude (Anthropic), Gemini (Google), Mistral and DeepSeek are families of LLMs.
No LLM is the best at everything: some excel at reasoning, others at writing, others at code, and open-source models can run locally on your hardware. That is why a multi-model platform, which lets you choose the right LLM for each task, is structurally more robust than a subscription to a single provider.
You paste a thirty-page contract and ask for the termination clauses: the model finds them and summarises them in a few seconds. The same kind of model writes an email, translates a manual or explains a table.
For any work on text: drafting, summarising, classifying, extracting, translating, answering questions. For exact calculations or up-to-date facts, it must be connected to your data and your tools.
Can an LLM be wrong?
Yes. It produces the most plausible text, not necessarily the most accurate, and it can make up an answer. Hence the value of connecting it to your documents, showing the sources and having a person validate what matters.
All AI models, deployed on your infrastructure, with your documents connected.