What is a large language model? The one-paragraph version
Agentric Team ·
The technology powering your AI receptionist is a large language model — LLM. Here's the honest, hype-free explanation of what that is.
The short version
A large language model is a system trained on a vast amount of text to predict what words should come next in a conversation. That sounds trivial; the result isn't — because to predict text well enough, the model has to internalise an enormous amount about language, reasoning, and how the world is described.
Ask it "the customer has a leaking tap, how should I respond?" and it doesn't look up an answer — it generates one, based on patterns learned from human knowledge, adjusted by your specific instructions (your knowledge base, your rules).
Why it changed software
Before LLMs, "understanding" customer language meant hand-built menus and keyword matching — the frustrating phone trees everyone hates. LLMs understand free-form language: "my boiler's making a weird noise" gets understood as a boiler fault enquiry without anyone programming that phrase. And they can decide — choosing to check availability, ask a clarifying question, or escalate, based on context.
What it's bad at (honestly)
- Facts it wasn't given — it can hallucinate (make things up). That's why your assistant is bounded by your knowledge base and tools, never free-styling facts.
- Perfect consistency — it's probabilistic, not a database. Good prompting + tool grounding keeps it reliable.
- Judgement beyond its brief — it won't diagnose a boiler fault; it captures symptoms for you.
Why it matters for reception work
Reception is language work: understanding what callers want, answering helpfully, and taking the right action. An LLM grounded in your business knowledge, bounded by your rules, and wired to your diary is — finally — software that can do that job.