DEFINITION
An AI model trained on huge amounts of text to understand and write language. In Clivio, an LLM summarizes your documents and answers the questions you ask about one of them.
How an LLM works
An LLM writes text by predicting what comes next, based on patterns learned from huge amounts of writing. It doesn’t look facts up in a database: it produces the most likely continuation of what it’s given. That makes it good at summarizing, rewording and answering questions about a text you hand it, and less reliable on facts it hasn’t been shown.
So the text you give it matters most. When the answer has to come from your lease or your invoice, the model is given that document first, and its answer should rest on it.
What it does in Clivio
Clivio reads your documents with Google’s Gemini models, through the Gemini API. When a file arrives, the model suggests a title, recognizes the type of document, writes a two- or three-sentence summary and picks out dates, amounts and names. In a document’s AI chat tab, you can then ask “When can I give notice?” and get an answer that cites the clause, with the passage highlighted in the PDF.
Each question counts as one AI action: 50 a month on the Free plan, 500 on Pro. Totals across your documents, such as what you spent with one supplier in 2026, are calculated on your data without the model, and cost no AI action.
Limits to keep in mind
An LLM can get things wrong, especially with a blurry photo or an unusual layout, and it can state a wrong answer with confidence, which is called a hallucination. Check anything that matters against the document itself. Clivio doesn’t sell your data or use it to train a model of its own, and the Trust page says where processing happens.
Frequently asked questions
Is ChatGPT an LLM?
ChatGPT is a chat app built on large language models made by OpenAI. Gemini, from Google, and Claude, from Anthropic, are other well-known LLMs.
Does Clivio train an AI on my documents?
No. Clivio doesn’t sell your data and doesn’t use it to train a model of its own. For the analysis, your documents’ content goes through the Google Gemini API, as described on the Trust page.