HOW TO PLAY
A mock interview scored like an RPG. The AI reads your job description, becomes the interviewer, and turns every answer into damage — in one direction or the other. Any field works: nursing, teaching, sales, trades, finance, engineering. The boss comes from your JD, not from a fixed question bank.
The loop
- Paste a job description and pick a difficulty in the lobby.
- The AI generates a boss persona from that JD and opens with a real question.
- You pick an action and answer. The judge grades it and assigns damage.
- Every third question is a quickfire round: four options, one clearly best.
- The boss asks a harder follow-up. Repeat until someone hits 0 HP.
- You get an XP score plus per-answer coaching notes — win or lose.
Actions
ANSWER
The standard attack. Answer the question straight. Graded on clarity, confidence, depth and relevance to the job description.
0–40 to the boss
STAR STRIKE
Commit to Situation, Task, Action, Result. Real structure lands a critical hit — a STAR Strike without a result is punished harder than a plain answer.
25–40 on a hit, brutal counter on a miss
CLARIFY
Ask a sharp question back instead of answering. A relevant clarifier chips the boss and buys you a better question. A lazy one still gets countered.
5–10 chip damage
QUICKFIRE (MULTIPLE CHOICE)
Every third question arrives as four answers instead of a blank box. One is clearly the strongest; the other three are mistakes candidates for this job really make. Pick one and it is graded like any other answer.
Best option crits, weak option gets countered
COFFEE
Two per run. The judge tells you what a great answer to the current question would cover. No damage either way — the question stays on the board.
0 damage, 2 uses
Damage table
| Rating | Earned when | Effect |
|---|---|---|
| CRITICAL_HIT | Specific, structured, evidence-backed, on target | 25–40 damage to the boss |
| HIT | Correct but shallow, vague, or missing impact | 10–20 damage to the boss |
| COUNTER_ATTACK | Weak, off-topic, evasive or empty | 15–30 damage to you |
Damage is clamped server-side, so the AI cannot invent a one-shot kill. Difficulty then scales it: hits and counters are multiplied before they reach your HP bar.
Difficulty
Intern run
Forgiving grader. Your hits land harder.
Normal
Fair interviewer. Fluff still gets punished.
Nightmare
Brutal grader. Counters hurt, hits chip.
Difficulty changes both how strictly the judge grades and how hard the numbers hit.
Boss roster
The Architecture Tyrant
Depth. Trade-offs, failure modes, and what you would do differently.
The Metric Gatekeeper
Business impact. Numbers, scope, and who was affected.
The Culture Auditor
Behavioural evidence. Real conflicts, real outcomes, no hypotheticals.
The Velocity Fanatic
Ownership and speed. What you shipped alone and how fast.
Tactics
- ▸Lead with the answer, then the evidence. The judge rewards the first sentence carrying the point.
- ▸Quantify everything you can — 'cut average patient wait time from 40 minutes to 12' or 'grew renewals 18%' beats 'improved things'.
- ▸Name the trade-off you rejected. It reads as judgement, not as recital.
- ▸If the question is genuinely ambiguous, CLARIFY once. Twice in a row reads as stalling.
- ▸Save your coffees for the question you would freeze on in a real interview.