Agent trajectory
The sequence of steps an agent takes to reach a result, such as which tools it called and in what order.
An agent trajectory is the sequence of actions an agent takes before it responds, such as checking history, looking up a policy or calling an API. Google's Agent Development Kit documentation recommends evaluating both the final output and the trajectory, comparing the actual steps with an expected one, because probabilistic agents make simple pass or fail tests insufficient. In finance the trajectory often matters as much as the answer: reading the wrong account or skipping an approval is a failure even if the final number is right.
Agent Minute explains this term on 15 March 2027.
Related terms
AI agentSoftware built on an AI model that pursues a set goal by planning steps and acting through tools or APIs, not only by answering.Tool useA model asking its host application to run a named function with structured arguments, then receiving the result.Model validationEvaluating whether a model performs as expected, including its reliability and its limitations.Ongoing monitoringContinuing checks that a model or agent is still performing as intended as data, use and conditions change.