AI Agent

Runs an autonomous AI agent that iterates through a goal using an OpenAI model, optionally calling tools, until it produces a final response.

Purpose

Use this task when a workflow needs to accomplish an open-ended objective that may require multiple reasoning steps or tool use rather than a single AI request. The agent loops internally — calling the language model, executing tools when needed, and refining its approach — until it reaches a final answer or exhausts its iteration limit. It is well suited for complex research, multi-step data gathering, or any scenario where the exact sequence of sub-tasks cannot be predetermined at design time.

Inputs

Field	Type	Required	Description
Agent Goal	Multi-line Text	Yes	The objective the agent should accomplish. Be specific about the desired outcome and any constraints.

Outputs

Name	Description
ConversationHistory	The full serialised message history from the agent session, including all model calls and tool interactions.
AgentOutput	The final response produced by the agent once it determined the goal was complete.
IterationsUsed	The number of reasoning iterations the agent performed before completing or reaching the iteration limit.