Term · Apps & software
LLM (large language model)
A large language model (LLM) is an AI model trained on very large amounts of text to predict and generate language. Modern LLMs can answer questions, summarise documents, draft texts, translate, classify messages, extract data from documents and hold conversations.
[ 01 ]
Explained
Business uses
- Customer support assistants answering from company knowledge. See AI assistants.
- Drafting replies for human agents.
- Summarising long conversations, documents and meetings.
- Extracting structured data from emails, invoices and forms.
- Classifying and routing incoming requests.
- Translating and localising content, with human review.
Limitations
- Invented answers: models can produce fluent but wrong statements. Grounding with RAG and guardrails reduces this. See AI hallucinations.
- Knowledge cut-off: models know only what was in their training data unless given current information.
- Language quality varies: strong in English, variable in less-resourced languages. See AI assistants in Azerbaijani.
- Consistency: the same question can produce different wording.
Costs and deployment
LLMs are usually accessed through providers' APIs and billed by the amount of text processed. Open models can be self-hosted, trading API costs for infrastructure and expertise.
Data considerations
Understand what data is sent to the provider, how it is stored and whether it may be used for training — and ensure compliance with personal data law. Send only what is needed.
[ 02 ]
FAQ
Frequently asked questions
Can an LLM replace our support team?
It can handle a share of routine questions and assist agents; complex and sensitive cases still need people.
Which LLM should we use?
Choose by testing on your tasks and languages, alongside cost, speed and data terms.