InstaSight v2

AI Academic Intelligence for Coaching Institutes

Move from spreadsheet-led operations to an AI-assisted workflow that improves student outcomes, reduces dropout risk, and makes parent reporting consistent.

Open LoginOpen Dashboard

ICP

50-500 students, 1-5 batches

Buyer

Institute owner/director

Primary User

Admin or academic coordinator

Secondary Stakeholder

Parents

MVP Scope (Phase 01)

  • •Institute onboarding + OTP verification + JWT auth
  • •Student/batch CRUD + CSV import
  • •Attendance tracking + threshold alerts
  • •Test result CSV upload + validation
  • •AI weak-area detection (Gemini)
  • •Dropout risk rule engine
  • •Dashboard + student performance card
  • •Parent progress PDF export

Out of Scope (MVP)

  • •Multi-admin RBAC
  • •Student self-portal
  • •Billing and feature gating
  • •Automated WhatsApp API alerts
  • •Cross-institute benchmarking

Architecture Blueprint

Browser (untrusted) → Next.js Frontend → FastAPI Backend → MongoDB Atlas / Gemini API / SMTP / PDF Service

Trust boundaries are enforced server-side, with tenant isolation via institute_id on all data reads/writes.

Core Collections

institutes, admins, otp_tokens, students, batches, attendance_records, tests, test_scores, ai_insights, risk_flags, report_exports, audit_logs.

API Contract (MVP)

Base path: /v1

  • •Auth: register, OTP verify, login, refresh, logout, forgot/reset password
  • •Students/Batches: CRUD + CSV validate/commit
  • •Attendance: mark, summary, exports
  • •Tests: CSV validate/commit, status
  • •Insights: student/class weak-area insights
  • •Risk: list/recompute/export
  • •Dashboard: summary + recent feed
  • •Reports: parent PDF generate/status

CSV + AI Pipeline

  1. Upload CSV
  2. Schema validation
  3. Row normalization
  4. Semantic validation
  5. Duplicate detection
  6. Validation preview
  7. Commit import
  8. Queue risk + AI jobs
  9. Persist insight outputs

Guardrails: strict JSON output contract for Gemini, retry with backoff, insight caching by test and prompt version.

6-Week Delivery Plan

  • •Week 1: foundation, auth, deploy baseline
  • •Week 2: student/batch CRUD + CSV import
  • •Week 3: test upload, AI insights, risk engine
  • •Week 4: dashboard + student performance card
  • •Week 5: attendance + PDF exports
  • •Week 6: hardening + pilot onboarding

Pilot Readiness and UAT

Readiness: env vars configured, health/readiness endpoints active, backup/restore tested.

Critical journey: register → import roster → upload tests → verify insights/risk → export parent report.

Observability baseline: structured logs, latency/error metrics, queue + AI failure alerts.