What is a large language model?
A Large Language Model (LLM) is a large language model that has learned linguistic relationships from extensive data. It can process text and generate responses. An LLM is the technical foundation of many AI assistants, but without additional connections it cannot access your internal systems.
AN EXAMPLE IN PRACTICE
Where you might encounter it
An assistant summarises a lengthy service report. The language model generates the summary; another component retrieves the approved report from your document system. Access permissions for the file are checked independently of the model.
Why does this matter for your business?
You need to distinguish what the model does from what the complete application requires. Data access, search, approvals and logging are all part of the solution around the model.
How does business knowledge get into the response?
One common option is RAG. Relevant sources are searched for and provided to the model as context. Alternatively, an adapted model may be suitable for specific tasks; assess this against your needs.
Frequently asked questions
Does an LLM automatically know my current prices?
No. Current prices must be provided from a reliable source or retrieved via an interface. Knowledge from training is no substitute for querying your inventory management system.
Do I need to train my own model?
For many initial projects, an existing model with suitable instructions and data access is sufficient. Custom adaptations are worthwhile only when tests show a specific additional benefit.
What to keep in mind
- Compare the model using typical tasks.
- Measure response quality and processing time.
- Control data access separately.
Further sources
From NetzTurm Decode. Clear explanations for your business.