---
id: "prereq-llm-vs-agent"
type: "prereq"
source_timestamps: ["¶3", "§ The Three Types of AI Agent Interactions"]
tags: ["ai-literacy"]
related: ["concept-brand-agents", "concept-consumer-agents", "framework-three-types-ai-interactions"]
reason: "Crucial for understanding the transition from passive research (using an LLM like a search engine) to active intermediation (delegating a purchase to an agent)."
sources: ["agentic"]
sourceVaultSlug: "hbr-seg-agentic"
originDay: 6
articleStem: "hbr-ext-18-preparing-brand-agentic-ai"
sourceUrl: "https://hbr.org/2026/03/preparing-your-brand-for-agentic-ai"
sourceTitle: "Preparing Your Brand for Agentic AI"
---
# Distinction between LLMs and AI Agents

**Prerequisite.** The source assumes the reader understands that an **LLM** (like GPT-4) is a foundational model that generates text, whereas an **Agent** (built on top of an LLM) has autonomy, memory, tool-use capabilities (browsing, API calls), and can execute multi-step tasks to achieve a goal.

**Why it matters.** This distinction is the hinge between *passive research* (using an LLM like a search engine) and *active intermediation* (delegating a purchase to an agent) — the very shift that produces [[concept-brand-agents]], [[concept-consumer-agents]], and the full taxonomy in [[framework-three-types-ai-interactions]].


## Related across articles
- [[prereq-agentic-vs-generative-ai-d6]]
- [[prereq-foundation-models]]
- [[prereq-agentic-ai-understanding-d16]]