Agent Terminology
Master the essential vocabulary and concepts in the agentic AI landscape
Your Progress
0 / 5 completedCommon Confusions
Some terms in agentic AI sound similar or get used interchangeably. Let's clarify the key differences.
🤔 Select a Common Confusion
🤖 Agent
What it is:
An autonomous system that uses an LLM to make decisions, take actions, and adapt based on outcomes.
Key characteristics:
- Makes decisions independently
- Uses tools to interact with environment
- Iterates based on feedback
- Maintains state across actions
Example:
Agent decides: "I need more info" → searches web → reads results → synthesizes answer → validates accuracy
💬 LLM Application
What it is:
A direct interface to an LLM with optional prompt engineering, but no autonomous decision-making.
Key characteristics:
- Responds to prompts directly
- No autonomous action-taking
- Stateless (unless manually managed)
- Human drives each interaction
Example:
User asks → LLM generates text response → done. No follow-up actions or tool usage.
🎯 Key Difference:
Agents have agency - they decide what to do next. LLM apps just respond. Think of it like the difference between an autopilot (agent) and a co-pilot suggestion system (LLM app).
💡 Pro Tip
When learning agentic AI terminology, focus on understanding relationships and distinctions rather than memorizing definitions. Ask yourself: "Is this about the mechanism (how), the capability (what), or the persistence (when)?" This framework helps clarify most confusions.