LLMs are built on the transformer architecture, which lets the model weigh how every word in the input relates to every other word. Training happens in stages: a pre-training phase where the model learns general language patterns from a huge text corpus, followed by fine-tuning and alignment steps that teach it to follow instructions and behave safely. The result is a general-purpose text engine rather than a program written for one specific task.
Because an LLM generates text statistically rather than looking answers up, it has no built-in knowledge of your business and no guarantee of factual accuracy. Production systems solve this by supplying the model with trusted source material at the moment of the question, a pattern known as retrieval-augmented generation, and by constraining behaviour with a system prompt.
Clerkzo uses an LLM as the language layer of its website chat widget, but never lets it answer from memory alone. Every reply is grounded in content crawled from the customer's own site, and guardrails keep the model on-topic and hand anything it cannot answer to a human.
Related terms
Browse all 142 terms- Document Object Model (DOM)The Document Object Model, or DOM, is the structured map a browser builds of your web page after loading it. Code can read and change that map, which is how anything on a page moves, updates or responds to a click.
- Model TemperatureModel temperature is a setting that controls how much randomness an AI model applies when choosing its next word. Lower values produce consistent, predictable output; higher values produce more varied and creative output.
- Transformer ModelA transformer model is a type of neural network architecture that processes text by weighing how much every part of the input relates to every other part. It is the design underpinning virtually all modern large language models.
- Natural Language ProcessingNatural language processing, usually shortened to NLP, is the field of computer science concerned with getting software to understand, interpret and produce human language in text or written form.