Concept guide · Agents, tools, and MCP
What is an AI agent?
A slower explanation for when the short interview answer is not enough. Start with the intuition, then connect it to the technical idea and the way an interviewer expects you to reason about it.
ML area
Generative AI → Agent fundamentals
Interview format
Foundations · Scenario
Question
GENAI-075
Start with the idea
In simple terms
An agent is a model-driven loop that decides what to do next, uses a tool or other action, observes the result, and continues until it can finish or needs help. The model suggests the next step; application code still controls what is allowed to happen.
Make it concrete
A support agent can classify an issue, search documentation, look up an account, draft a response, and pause for approval before sending anything externally.
Build the answer
How to approach it
- 1Define the goal, available actions, and what counts as completion.
- 2Separate model decisions from application-enforced permissions and side effects.
- 3Compare an agent with a fixed workflow using quality, cost, latency, and failure evidence.
Go one level deeper
The technical explanation
The useful distinction is between a model call and a system with state, actions, observations, and control flow. An agent needs a goal, bounded tools, a working state, stopping conditions, retries, permissions, and an evaluation target. If the steps are known in advance, a fixed workflow is often safer and cheaper; autonomy is justified only when the path genuinely depends on intermediate results.
Say this in the interview
An agent is a model-driven loop that observes a goal and context, chooses an action or tool, receives a result, and decides whether to continue or finish. The application still owns state, permissions, limits, execution, and validation; autonomy is not a substitute for engineering controls.
Keep preparing