Difficulty isn't part of the blueprint — it's a choice you make each time you generate questions, on the role's Generate screen. Every role can have up to three separate question sets, one per difficulty, each with its own version history.
Difficulty controls how many questions get generated and how deep they go:
Phone screen — a high-level fit check, 3–5 questions. Good for an initial screen before a deeper technical round.
Technical screen — validates core competency in more depth, 8–12 questions.
In-depth — detailed technical and behavioral probing, 12–20 questions. Best for final rounds or senior/staff-level roles.
The Generate screen suggests a starting question mix for whichever difficulty tab you're on — pulled from the role's seniority the first time, or from your last generation's own counts if you're regenerating. You can always adjust the suggested amounts before generating.
If a difficulty tab already has questions generated, clicking Generate again just reuses them instantly — no new AI call, no wait. Regenerate is separate and deliberate: it creates a new version and counts against your monthly generation limit, so it asks you to confirm first.