Agent Loop
An agent loop is a repeated cycle. The AI thinks about the goal, plans an action, acts, observes the result, and decides what to do next.
Loop Playground
Step through the loop
Press Next Step to move around the loop. Watch how each observation changes what the agent does next.
Goal
Current stage
Goal complete
Ready
Press “Next Step” to start the loop.
Step 0 of 6
Loop stages
The stages of the loop
Each turn of the loop moves through the same simple stages.
Why the loop matters
One answer, or many small steps?
The loop helps the agent handle tasks where the first action may not be enough.
Without loop
The agent gives one answer and stops.
Ask
Answer
Stop
With loop
The agent checks progress and improves the result step by step.
Act
Observe
Adjust
Repeat
🔑
Key takeaways
- • Agentic AI often works in a loop.
- • The loop helps the agent move toward a goal.
- • Each action creates an observation.
- • Observations help decide the next step.
- • The loop ends when the task is complete.
Summary
The agent loop helps an AI system keep working toward a goal. It acts, observes what happened, adjusts the next step, and repeats until the task is complete.