The expectation is that a simple prompt from a user will enable an agent to classify intent, break down the goal into multiple steps and complete the task, whether it includes internet searches, authentication into multiple tools or learning from past repeat behaviors.
You are viewing a single comment's thread from: