Question

A product team wants to build a feature that investigates customer complaints. Each complaint requires an unpredictable number of steps: reading the ticket, searching internal knowledge bases, querying order histories, and sometimes escalating to check shipping partner APIs — but the exact sequence and which tools are needed vary widely from case to case and cannot be known in advance. The team asks whether to implement this as a fixed workflow or as an agent. Which approach best fits this task, and why?

Accepted Answer

An agent, because the task requires the model to dynamically direct its own control flow and choose which tools to use based on what it discovers at each step.

Answer

A fixed workflow, because chaining a predetermined set of prompts is always more reliable than letting the model decide anything.

Answer

A fixed workflow using parallelization, because running all possible tool calls simultaneously guarantees every case is covered.

Answer

An agent, but only because agents are cheaper and faster than workflows for every task type.