Agent Loop
The iterative cycle of an AI agent: Observe → Think → Act → Evaluate result → Repeat until goal is reached.
The agent loop is the iterative observe-think-act cycle that drives AI agents – the fundamental pattern behind autonomous task execution.
Explanation
The agent loop is the heart of every AI agent. Per iteration, the LLM decides whether to call a tool, ask a question, or deliver the final answer. Good loops have exit conditions, max iterations, and backoff strategies.
Marketing Relevance
Understanding the agent loop is critical for designing and debugging AI agents and optimizing their costs.
Common Pitfalls
Infinite loops without exit condition. Too many tool calls per iteration. Missing backpressure on long runs.
Origin & History
The concept is based on the OODA loop (Boyd, 1976) and was applied to LLM agents through ReAct (Yao et al., 2022).
Comparisons & Differences
Agent Loop vs. Chain-of-Thought
CoT is one-time step-by-step thinking. Agent loop is an iterative cycle with tool execution and feedback.