Flashcard: what is an AI agent, in one sentence?

Attempt it

Your answer

**An AI agent** is a language model running inside a loop that can use tools. The model reads a task, decides what to do next, and either emits a final answer or asks for a tool to be called. If it asks for a tool, the runtime around the model actually calls that tool, captures the result, and feeds it back. Then the model gets another turn. This back and forth keeps going until the model says it is done or the runtime hits a stopping condition. The one sentence version is: an agent is an LLM-driven loop that observes, reasons, acts, observes again, and repeats until the task is done. Two parts make this work. The model is the brain that decides. The runtime is the body that actually acts on the world by calling tools, querying databases, running code, or sending messages. Without the loop you only have a chatbot. Without the tools you have a model that can think but cannot do anything. Without the model you have plumbing with nothing to drive it. All three together are what we call an agent. Frameworks like LangGraph, CrewAI, AutoGen, and the OpenAI Agents SDK each provide a runtime that runs this loop for you so you do not have to write the orchestration from scratch.

**An AI agent** is an LLM-driven loop that takes a task, observes the current state, reasons about the next step, calls a tool or external system, reads the result back as a new observation, and repeats until it produces a final answer or hits a stopping condition. The one-sentence definition compresses three ideas that have to all be present. First, there is a language model doing the deciding. Second, there is a tool layer the model can invoke to act on the world beyond text generation. Third, there is a runtime that actually drives the loop, parses the model's tool requests, executes them, and feeds the observations back in. The model is the brain. It only ever produces tokens, and on each turn those tokens are either a final answer or a structured request to call a specific tool with specific arguments. The runtime is everything else. It owns the tool registry, the conversation state, the step and cost budgets, and the stopping conditions. This is what separates an agent from a one-shot completion. A chatbot answers once and stops. An agent keeps going across many turns, with each turn shaped by what the previous tool calls returned. That iterative, observation-driven control flow is the whole point. Production agent runtimes include LangGraph, CrewAI, AutoGen, and the OpenAI Agents SDK; all of them implement the same observe, reason, act, repeat loop.

Quick reference

TL;DR

An agent is an LLM in a loop with tools: it observes, reasons, acts, reads results, and repeats until it produces a final answer or a stopping condition fires.

Memory aid

Easy to grasp

Imagine giving a very fast intern a goal, a phone, a calculator, and a notepad. The intern reads the goal, decides one next step, picks up a tool to do that step, writes down what they learned, and then looks at the notepad again to decide what to do next. They keep going until they have an answer or they run out of time. An AI agent works the same way. The language model is the intern doing the thinking. The runtime is the office around the intern: it actually picks up the phone, runs the calculator, writes things down on the notepad, and watches the clock. The two together, taking turns over and over, are what people mean when they say agent.

Key concepts

Agent Loop Tool Use Agent Runtime

Detailed answer & concept explanation

~7 min read

Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example. Click to expand.

Everything you need to truly understand this topic: intuition, mechanics, step by step explanation, code, formulas, and worked example.

Quick scan · 2–4 min

Everything important, quickly.

5 min: state the one-sentence definition, unpack the observe, reason, act, repeat loop, separate the model's role from the runtime's role, contrast against a one-shot completion and a fixed chain, and name two production runtimes.

Real-world examples

Real products, models, and research that use this idea.

Anthropic's computer use feature: Claude Opus 4.7 decides the next click or keystroke, the harness takes the screenshot and executes the action, then loops the new screen back in.
Cursor and Cline coding agents: the model proposes an edit, the editor runtime applies the diff and runs tests, and the test output becomes the next observation.
LangGraph state graphs and the OpenAI Agents SDK are runtimes that drive this observe reason act loop in production deployments.
Devin and similar autonomous coding agents run the same loop across hours, with tool calls to a sandboxed shell, a browser, and a file system.

Interviewer follow-ups

What an interviewer would ask next. Try answering before peeking at the approach.

QIf the model only emits text, what actually invokes a tool and how does that result get back into the prompt?

QHow is an agent different from a LangChain-style chain that calls the LLM several times?

QWhat stops an agent from looping forever?

Don't say this

Red flags and common mistakes that signal junior thinking. Click to expand.

Most common mistake

Calling a single LLM completion an agent. A chatbot answer is not an agent; an agent requires a loop, tools, and a runtime that drives the loop until a stopping condition fires.

60-second cheat sheet

The night-before-the-interview bullets. Scan these on the way to the call.

References

Primary sources. Skim if you want the original framing.