On this page
How does a large language model work?
During training, an LLM learns statistical patterns in language, how words, ideas and sentences relate, from a very large body of text. When you give it a prompt, it predicts a fitting response one piece at a time. That’s what lets it understand a question however it’s phrased and reply fluently in many languages.
What are LLMs good at?
- Understanding intent in messy, informal questions.
- Writing clear, on-brand text such as replies, product descriptions and blog posts.
- Summarising and translating.
- Adapting tone for different audiences.
What are their limits?
An LLM doesn’t inherently know facts about your business, and its general knowledge has a cut-off date. It can also produce plausible but incorrect statements, known as AI hallucination. That’s why customer-facing uses pair LLMs with retrieval-augmented generation, clear boundaries and human oversight.
How Yep AI uses it
Yep AI Agents use language models to understand customers in 90+ languages and write in your brand voice, while answers are grounded in your Shopify catalogue, policies and order data. The AI Marketing Agent, Marketing Specialist, uses the same capability to write SEO blogs, product copy and social posts: see ecommerce marketing.
