A developer is designing a GitHub Copilot agent that will automatically triage incoming issues and assign them to team members. Which sequence correctly describes the fundamental steps an agentic AI system performs?
Select an answer to reveal the explanation.
Short Explanation and Infographic
Think of an agent like a contractor who first surveys the job site (perceive), figures out what needs doing (reason), writes up a work order (plan), then swings the hammer (execute). Skipping any step leads to rework or disaster. The correct agentic loop is perceive → reason → plan → execute.
Full explanation below image
Full Explanation
Agentic AI systems follow a structured loop: Perceive (gather inputs from the environment — issues, code, context), Reason (analyze what the inputs mean relative to the agent's goal), Plan (determine a sequence of actions to take), and Execute (carry out those actions and observe the results). This cycle may repeat many times in a single task.
Why B is correct: This is the canonical agent loop described in agentic system design. Separating perception from reasoning from planning from execution gives developers control points at each stage and prevents the agent from acting blindly.
Why A is wrong: Executing before perceiving means the agent acts without knowing the current state. This is a classic anti-pattern that causes agents to make changes based on stale or incomplete information.
Why C is wrong: Executing all available actions simultaneously without planning leads to conflicts, wasted compute, and irreversible mistakes. Reasoning must come before execution, and execution should be selective.
Why D is wrong: Storing instructions and executing immediately skips the planning and reasoning stages entirely. This removes the agent's ability to validate its approach, check for conflicts, or adapt to context — all critical for safe agentic operation.