Should this be a trained model or an LLM-based agent — and where is the boundary between the two?
Answer it out loud before you open anything. The value of the flags below is in comparing them to what you actually said — including whether you asked about the data before naming a model.
An operations team wants to "use AI" to triage incoming supplier invoices: extract fields, flag likely duplicates and anomalies, and route exceptions to a human. Volume is 40,000 invoices a month. One proposal is an agent with tools that reads each invoice; another is a set of trained models. A third proposal is "both, somehow".
What it is really testing
Whether the candidate can decompose a vague ask into sub-problems and assign each to the right kind of system from its properties — volume, latency, cost per item, need for a probability, availability of labels, stability of the task, need for judgement — rather than from fashion in either direction.