# 2. AI Mentor Mode Settings

User Type: **Super Administrator**
Source: *Mi Digital Academy - Education CRM Features Document*

---

## 2. AI Mentor Mode Settings

### 2.1 Mentorship Style Configuration
**What it does:** Configures the AI mentor's mentorship style: the communication tone, the check-in frequency, the goal-setting approach, and the feedback style.

**Sub-features:**
- Communication tone (encouraging, structured, accountability-focused)
- Check-in frequency (daily, weekly, per milestone)
- Goal-setting approach (SMART goals, milestone-based)
- Feedback style (immediate, batched, reflective)
- Mentor persona selection (per grade band)
- Available on web
- Event logging (action performed)
- Audit logging of mentorship style configuration

**Super Admin User Journey:**
1. Super Admin opens AI Feature Configuration → AI Mentor Mode → Mentorship Style.
2. Sets the communication tone (encouraging, structured, accountability-focused).
3. Configures the check-in frequency (daily, weekly, per milestone).
4. Sets the goal-setting approach (SMART goals, milestone-based) and the feedback style.
5. Selects the mentor persona per grade band.
6. Saves; the style applies to all AI mentor sessions.
7. Each configuration change is event-logged and audit-logged.

**Rules & Edge Cases:**
- A grade band with no assigned persona falls back to the default persona.
- The check-in frequency is the maximum cadence; the mentor does not check in more often than configured.
- The feedback style determines when feedback is delivered; batched feedback is delivered at the configured interval.
- Configuration changes are audit-logged with the setting, the change, and the timestamp.

### 2.2 Goal Tracking Settings
**What it does:** Configures the AI mentor's goal tracking: the goal types, the goal milestones, the progress measurement, and the goal review cadence.

**Sub-features:**
- Goal types (exam target, skill mastery, habit goal)
- Goal milestone definition (auto-generated, manual)
- Progress measurement (metric per goal type)
- Goal review cadence (weekly, bi-weekly, monthly)
- Goal adjustment rules (re-target on sustained under/over-performance)
- Available on web
- Event logging (action performed)
- Audit logging of goal tracking settings

**Super Admin User Journey:**
1. Super Admin opens AI Feature Configuration → AI Mentor Mode → Goal Tracking.
2. Sets the goal types (exam target, skill mastery, habit goal).
3. Configures the goal milestone definition (auto-generated or manual).
4. Sets the progress measurement metric per goal type.
5. Defines the goal review cadence and the adjustment rules.
6. Saves; the settings apply to all mentor-tracked goals.
7. Each configuration change is event-logged and audit-logged.

**Rules & Edge Cases:**
- A goal with no measurable metric is not trackable; the mentor prompts for a measurable target.
- A goal re-target is proposed by the mentor and confirmed by the student; it is not applied silently.
- Goal progress is measured from the configured metric; there is no manual progress override.
- Configuration changes are audit-logged with the setting, the change, and the timestamp.

### 2.3 Study Habit Analysis Parameters
**What it does:** Configures the AI mentor's study habit analysis: the habits tracked, the analysis window, the pattern detection, and the habit recommendation rules.

**Sub-features:**
- Habits tracked (study time, consistency, subject balance, peak hours)
- Analysis window (7-day, 30-day rolling)
- Pattern detection (consistency, procrastination, cramming)
- Habit recommendation rules (improvement suggestions per pattern)
- Habit report cadence (weekly, monthly)
- Available on web
- Event logging (action performed)
- Audit logging of study habit analysis parameters

**Super Admin User Journey:**
1. Super Admin opens AI Feature Configuration → AI Mentor Mode → Study Habit Analysis.
2. Sets the habits tracked (study time, consistency, subject balance, peak hours).
3. Configures the analysis window (7-day, 30-day rolling).
4. Defines the pattern detection (consistency, procrastination, cramming).
5. Sets the habit recommendation rules and the report cadence.
6. Saves; the analysis applies to all mentor-tracked students.
7. Each configuration change is event-logged and audit-logged.

**Rules & Edge Cases:**
- A student with insufficient activity in the window is not analyzed (no false patterns from sparse data).
- A detected pattern produces at most one recommendation per pattern per report cadence (no repetition).
- The analysis window is rolling; it is not reset by the report cadence.
- Configuration changes are audit-logged with the parameter, the change, and the timestamp.

### 2.4 Emotional Intelligence Thresholds
**What it does:** Configures the AI mentor's emotional intelligence thresholds: the sentiment detection sensitivity, the frustration indicator, and the motivation level assessment.

**Sub-features:**
- Sentiment detection sensitivity (low, medium, high)
- Frustration indicator (repeated errors, negative language, session abandonment)
- Motivation level assessment (engagement signals, goal progress)
- Threshold per indicator (trigger level)
- Threshold per grade band (age-appropriate sensitivity)
- Available on web
- Event logging (action performed)
- Audit logging of emotional intelligence thresholds

**Super Admin User Journey:**
1. Super Admin opens AI Feature Configuration → AI Mentor Mode → Emotional Intelligence.
2. Sets the sentiment detection sensitivity (low, medium, high).
3. Configures the frustration indicator (repeated errors, negative language, session abandonment).
4. Sets the motivation level assessment (engagement signals, goal progress).
5. Defines the threshold per indicator and per grade band.
6. Saves; the thresholds apply to all mentor sessions.
7. Each configuration change is event-logged and audit-logged.

**Rules & Edge Cases:**
- A threshold trigger produces at most one intervention per indicator per session (no repeated interventions).
- A grade band with no specific threshold uses the global threshold.
- The sentiment detection operates on the student's text input only; it does not infer from silence alone.
- Configuration changes are audit-logged with the threshold, the change, and the timestamp.

### 2.5 Human Support Trigger Rules
**What it does:** Defines the rules for triggering human support from the AI mentor: the trigger conditions, the support routing, and the escalation context.

**Sub-features:**
- Trigger conditions (sustained low motivation, repeated frustration, student request)
- Support routing (counselor, parent notification, academic advisor)
- Escalation context (session summary, indicator values, student history)
- Trigger cooldown (minimum interval between triggers per student)
- Support availability window and fallback
- Available on web
- Event logging (action performed)
- Audit logging of human support trigger rules

**Super Admin User Journey:**
1. Super Admin opens AI Feature Configuration → AI Mentor Mode → Human Support Triggers.
2. Sets the trigger conditions (sustained low motivation, repeated frustration, student request).
3. Configures the support routing (counselor, parent notification, academic advisor).
4. Defines the escalation context (session summary, indicator values, student history).
5. Sets the trigger cooldown and the availability window with fallback.
6. Saves; the rules apply to all mentor sessions.
7. Each configuration change is event-logged and audit-logged.

**Rules & Edge Cases:**
- A student-requested support trigger always succeeds (routed or ticketed); it is never blocked by the cooldown.
- A trigger within the cooldown window is suppressed for indicator-based triggers but logged as suppressed.
- The escalation context is generated automatically; the human support recipient sees it before responding.
- Configuration changes are audit-logged with the rule, the change, and the timestamp.