SkilsMVP
BootcampInterview prep

AI Mock Interview — Spec

AI Mock Interview — Spec

This document specifies how the app runs an AI-driven mock interview using the drill banks in this folder (behavioral.md, scenario-roleplay.md, cs-knowledge.md, case-exercise.md, motivation-fit.md). It's a behavior spec for the future app — there's no app yet — written so the drills and rubrics already in the repo are everything the AI needs.

Goal

Give the learner realistic, repeatable interview practice with structured feedback, approximating a human CSM hiring manager within the self-serve model. This is what turns "I read about the CSM interview" (Topic 10) into "I've practiced it dozens of times" — including the role-play and scenario rounds that catch first-time CSM candidates off guard.

Session flow

  1. Choose a mode: a single question, or a full mock (e.g., 2 behavioral + 1 scenario role-play + 1 cs-knowledge + 1 case-exercise, opening with a motivation-fit warm-up), optionally filtered by type or difficulty.
  2. Ask a question. The AI poses a drill's prompt verbatim and waits — it does not reveal the Framework, Model answer, or Rubric.
  3. Let the learner answer, ideally thinking aloud (typed or spoken). The AI may give one light nudge if the learner is stuck or silent (e.g., "What outcome is this customer actually trying to reach?"), mirroring a real interviewer — but never feeds the answer.
  4. Optionally probe with a follow-up a real interviewer would ask ("How would you know it worked?" / "What would you do if they still churned?").
  5. Score and give feedback against the drill's Rubric when the learner finishes.

Role-play questions (scenario-roleplay.md)

These drills are live role-plays: the AI stays in character as the customer or stakeholder (e.g., a frustrated admin threatening to cancel, an exec on a QBR call, a champion who just left). It plays the role the drill specifies, responds in character to what the learner says, and pushes back the way that persona would — it does not narrate the answer or break character to coach mid-scene. After the learner brings the conversation to a close (or taps out), the AI drops the persona and scores the exchange against the drill's Rubric: empathy and acknowledgment first, getting to the underlying problem, owning what's ownable without over-promising, and steering to a concrete next step.

Scoring & feedback

  • Evaluate the answer against the drill's Rubric signals and the Framework structure. Produce: an overall rating (e.g., Needs work / Solid / Strong), per-criterion notes (what was good, what was missing), and 2–3 concrete, actionable improvements.
  • Only after scoring, optionally show the Model answer as a reference — never before.
  • Be specific and kind: cite what the learner actually said. The aim is improvement, not a grade.
  • Watch for the classic misses the rubrics encode, by drill type:
    • behavioral — not using STAR, vague stories with no concrete result, blaming the customer or colleague, no lesson learned, claiming credit for the team's work instead of centering their own actions.
    • scenario-roleplay — jumping to a solution before acknowledging the customer's frustration, getting defensive, over-promising (refunds, features, dates) to make the tension stop, or failing to land a clear next step and owner.
    • cs-knowledge — fuzzy or wrong definitions of core metrics (churn vs. retention, NRR/GRR, health score, NPS), confusing CSM with Support/Sales/AM, or reciting a term without explaining why it matters to the business.
    • case-exercise — no structure, no clarifying questions, diving into tactics before defining the customer's desired outcome and the success metric, ignoring risk/health signals, or proposing a plan with no way to measure whether it worked.
    • motivation-fit — generic answers ("I like helping people"), no real knowledge of this company/product, no tie to genuine transferable strengths from a prior career (account management, support, teaching, hospitality), or sounding rehearsed and hollow.

Repeat & improve loop

  • Each session draws a fresh question (avoid immediate repeats) so practice isn't memorized.
  • Track which rubric criteria the learner repeatedly misses across sessions and surface a "focus next on…" summary (e.g., "you often jump to a fix before acknowledging the customer," or "you skip defining a success metric in case exercises").
  • Tie into the program (../role.md): mock-interview checkpoints recommend specific drill types by phase (e.g., motivation-fit and behavioral early; cs-knowledge once Topic 7 metrics are covered; scenario-roleplay and case-exercise before the job hunt).

Guardrails

  • The AI is a coach, not a gatekeeper — encourage retrying.
  • Never expose the rubric/model answer before the learner attempts the question.
  • Keep feedback grounded in the rubric; don't invent criteria.
  • In role-plays, stay in character until the scene ends; switch to coach voice only to score.

Inputs the AI uses (already in the repo)

  • The drill prompt (the question or role-play setup), its Framework (ideal structure), Model answer (reference), and Rubric (evaluation signals) — from the five drill files in this folder.
  • No additional data needed; the content is the spec.