AI agents
Words made simple
A few useful meanings before you read the explanation.
- Agent
- An application that uses a model to choose its next action within defined limits. It can use tools and continue through several steps.
- Token
- A small piece of text processed by a model. It may be a whole word, part of a word, or punctuation.
- Evaluation
- Checking results against examples and clear criteria to find whether the system works well and where it fails.
Beginner — Agent versus chatbot?
- A chatbot is a conversational interface.
- An agent uses a model to choose bounded actions and observes results toward a goal.
- A chatbot can use an agent, but a one-call conversation does not require one.
Intermediate — What is an agent loop?
- The model proposes a step, the application validates and executes a permitted action, and the result becomes an observation.
- The loop repeats only while task progress and step/time/token limits allow it.
Senior — When are multiple agents justified?
- When distinct scopes or specialties show measurable benefit over one agent or a fixed workflow.
- I require typed responsibilities, limited delegation, shared budgets, consistency checks, and outcome evaluation—not cosmetic role-play.
Scenario — An agent keeps repeating search. How do you fix it?
- I inspect whether the goal and completion conditions are clear and whether tool results are usable.
- I add repeated-query/progress detection and strict budgets, then clarify or escalate when evidence remains insufficient rather than searching indefinitely.
Can you explain it without the notes?
AI agents
Use your own words. Explain an example and one limitation before checking the explanation.
Check your explanation
Beginner — Agent versus chatbot?
Turn this concept into a clear answer.
Practice this conceptAdapted from 13_AI.MD · “29.8 AI agents”. Examples are study fragments; framework behavior can depend on your version.