LLM (Large Language Model)

A transformer-based model trained on language data to predict and generate text.

An LLM is a deep neural network (typically a transformer) trained on text to predict the next token. Modern LLMs are also capable of code generation, structured reasoning, tool use, and multimodal understanding. We treat the LLM as a component, not a product: production value comes from how you wrap it with retrieval, evaluation, and controls.

When it matters

Always — the term is so generic it appears in nearly every conversation. Use it precisely: an LLM is the model, not the workflow, not the agent, not the product.

Real example

An AI-native workflow has several LLM calls per case (classification, retrieval-rerank, draft-generation, review-validation) — not 'an LLM'. The workflow is the product; the LLM is the engine.

KPIs to watch

LLM cost per case (full inference cost summed across calls), LLM-call success rate (>99.5%), LLM provider redundancy (at least 2 providers wired up).

Related terms

Foundation model

A large model pre-trained on broad data, then adapted to many downstream tasks.

Transformer

The neural network architecture that powers modern LLMs, based on self-attention.

Context window

The maximum number of tokens a model can process in a single request.

Frontier model

The leading-edge foundation models with the highest reasoning, coding, and multimodal capabilities.

See it in action

We use this every week

Book a 30-min call and we'll walk you through how LLM (Large Language Model) shows up in a real engagement we're running.

Book a 30-min call