Understanding AI Agents

Labels

What fundamentally distinguishes an AI agent from a standard chatbot?

2 months ago

In the agentic loop, what happens immediately after an agent takes an action (e.g., calls a tool)?

2 months ago

Why do AI agents need external tools like web search or code execution, rather than relying solely on their training data?

2 months ago

Which planning technique has an agent write out its reasoning step-by-step before producing a final answer, improving accuracy on complex tasks?

2 months ago

An agent is given the task: 'Research the topic, write a draft, then email it.' How should a well-designed agent approach this?

2 months ago

How does an agent typically decide it has completed its task and should stop the agentic loop?

2 months ago

A tool call fails with a network timeout. Which error-recovery strategy best preserves overall task progress?

2 months ago

How does an agent maintain state across multiple steps within a single task session?

2 months ago

In a multi-agent system, what is the primary role of an orchestrator agent?

2 months ago

Which failure mode describes an agent that repeatedly calls the same tool with the same inputs without making progress toward its goal?

2 months ago

At which point is it most appropriate to add a human-in-the-loop checkpoint in an agentic workflow?

2 months ago

When is a hardcoded pipeline a better choice than an agentic system for a given task?

2 months ago

An agent needs to recall a document it summarised three weeks ago — far beyond its context window. Which memory type is most appropriate?

2 months ago

Why should an AI agent be given the minimum set of permissions necessary to complete its task?

2 months ago

What is a 'hallucinated tool call' and why is it a significant risk in agentic systems?

2 months ago

What is a key advantage of splitting a complex task across multiple specialized agents instead of using a single agent?

2 months ago

Which of the following best describes how a tool result is fed back to an agent after a tool call?

2 months ago

An agent is tasked with sending 1,000 marketing emails. A user asks it to 'make them more personal.' The agent has already sent 200. What is the safest course of action?

2 months ago

What is the difference between in-context memory and external memory in an AI agent?

2 months ago

Which of these best illustrates the 'lost context' failure mode in a long-running agent?

2 months ago