AI Intelligence Suite
AI Assistant Configuration & Model Settings
How school directors configure AI providers, OpenRouter API keys, model selections, token budgets, and autonomous autopilot switches.
AI Assistant Configuration & Model Settings
School Directors and System Administrators can configure, customize, and monitor the AI intelligence engine across the school from Director Portal > School Settings > AI Configuration.
1. Provider & LLM Model Setup
YeneSchool natively supports OpenRouter-compatible LLM endpoints as well as direct provider integrations:
┌────────────────────────────────────────────────────────────────────────┐
│ ⚙️ AI Engine Configuration │
├────────────────────────────────────────────────────────────────────────┤
│ Provider: [OpenRouter (Recommended) ▼] │
│ API Key: [••••••••••••••••••••••••••••••••] [Test Key] │
│ Primary Model: [anthropic/claude-3.5-sonnet ▼] │
│ Fast / Utility Model: [google/gemini-2.0-flash ▼] │
│ Temperature: [0.3 (Deterministic / Pedagogical) ▼] │
└────────────────────────────────────────────────────────────────────────┘
- Navigate to Director Portal > School Settings > AI Configuration.
- Enter your OpenRouter API Key or custom gateway URL.
- Select your preferred models:
- Primary Model: Used for in-depth 5E lesson planning, exam question generation, and complex risk diagnostics.
- Fast Model: Used for real-time streaming chat, quick translations, and morning executive summaries.
- Click Test Connection to verify API key validity and quota status.
- Click Save Configuration.
2. Autonomous Autopilot Switches
Enable or disable specific background agents to suit your school's operating preferences:
| Autopilot Switch | Default | Operational Impact |
|---|---|---|
AI_MORNING_BRIEFING_ENABLED | true | Synthesizes daily 06:30 AM executive briefing card for School Director. |
AI_PROACTIVE_TIMETABLE_WATCHDOG | true | Detects teacher absences and automatically negotiates Telegram substitute swaps. |
AI_TIMETABLE_AUTO_APPROVE | false | When true, automatically commits substitute periods without requiring manual Director sign-off. |
AI_EMPATHETIC_PARENT_DIALOGUE | true | Formats grade drop and attendance alerts with warm pedagogical encouragement. |
AI_EARLY_WARNING_CRON | true | Runs continuous student dropout and academic risk score evaluations. |
3. Token Budgets & Rate Limiting
To prevent unexpected API costs and ensure equitable access:
- Daily Token Limits per Role: Set maximum generation limits for teachers, students, and administrative staff.
- Context Window Cache: Repeated queries against the same school curriculum units are cached in Redis to minimize token consumption by up to 60%.
Next Steps in AI Intelligence
- Interactive AI Assistant & Copilot — Real-time chat assistant and prompt library
- Autonomous Background Autopilots — Risk detection and daily briefings
- Timetable Substitution Watchdog — Automated absence matching and Telegram period swaps
- Empathetic Parent Dialogue Engine — Automated empathetic SMS alerts