Learn · AI Agents
AI agents & tool use
An agent doesn't answer in one shot. It reasons, takes an action (calling a tool), reads the result, then loops — until it has enough to answer. This is the ReAct loop. Pick a task and step through it.
Tools the agent can call
Pick a task:
Task
Pick a task above to begin.
The agent's reasoning trace will appear here, step by step.
Why agents?
A plain language model only predicts text. An agent wraps the model in a loop and gives it tools — search, a calculator, a database, an API. The model decides which tool to call and with what input, reads the result, and keeps going until the task is done. That's how AI moves from answering questions to actually getting things done — the shift behind agentic workflows.
The reasoning traces here are illustrative and scripted to show the shape of the ReAct loop clearly. A real agent generates each thought and tool call live from the model, and tool results come from real systems. Client names and figures shown are fictional examples.