Do not begin with “which model do you want to use?” Understand the work first, then decide where AI belongs.

User and trigger

  1. Who performs this work every day?
  2. What event triggers the task?
  3. What format is the input, and where does it come from?
  4. Who receives the result, and what later decision does it affect?

Current workflow and cost

  1. How long does one complete cycle take today?
  2. Which step is slowest, most error-prone, or most dependent on experience?
  3. What are the normal and peak volumes?
  4. What is the real impact of a wrong result?

Data and systems

  1. May the data enter a model, and what must be redacted?
  2. Which internal systems and tools must be connected?
  3. Which actions may run automatically, and which require approval?
  4. What logging, audit, permission, and retention rules apply?

Success criteria

  1. Which three metrics must improve for the business team to call the project valuable?
  2. Who accepts the result, and who decides whether usage expands?
  3. How quickly can the team collect the first real feedback?

After the interview, write a one-page project judgment before writing code: target user, current workflow, critical constraints, minimum loop, success metrics, and stop condition.