Who It's For
Studeia serves four types of institutions:
| Vertical | Typical Use Case | Priority Features |
|---|---|---|
| K-12 Schools | Elementary and high school, BNCC (Brazil's national curriculum framework) alignment, parent portal | AI-narrative parent portal, gamification, minor monitoring, LGPD compliance |
| ENEM/College Test-Prep Courses | Live mock exams, personalized AI tutor, analytics | Psychometric quiz analytics, RAG-powered AI tutor, mock exams |
| Universities | Hybrid courses, LTI with existing Canvas/Moodle, gradebook | LTI 1.3 (tool provider), multi-criteria rubrics, AGS grade passback |
| Corporate Training | L&D, compliance, SSO with Okta/Azure AD | SAML/OIDC, SCIM 2.0, verifiable certificates (OB 3.0), public API |
Architecture in One Sentence
Next.js 15 (App Router) + PostgreSQL (Supabase) + Redis + Prisma 7 + multi-provider LLM Router (Claude → GPT → Grok → Gemini fallback) + Voyage AI embeddings (RAG) + Stripe/Asaas billing, deployed in Docker behind Caddy with on-demand TLS (automatic Let's Encrypt).
Core Features
1. True Multi-Tenancy with Full White-Label
Each institution (Tenant) has data isolation at the database level (tenantId filter on all queries + RLS in Supabase). White-label covers:
- Colors, logo, favicon, and font
- Custom domain (with automatic TLS via Let's Encrypt)
- 9 visual themes (6 dark glass + 3 light solid/material)
- Institutional email with branded templates
- Option to fully disable the Studeia brand
Details in Multi-tenancy and White-label.
2. Multi-Agent AI Tutor with Per-Tenant RAG
Studeia's AI tutor is not a ChatGPT wrapper. It's a multi-agent pipeline:
Student message
↓
StudentModelService → loads ConceptMastery + misconceptions + quiz performance
↓
RetrievalAgent → tenant-scoped RAG search, boosted by weak quiz areas
↓
PedagogicalAgent → selects strategy (direct_instruction, scaffolding, socratic,
guided_practice, challenge) based on mastery
↓
Orchestrator → builds enriched prompt
↓
LLM (Claude → GPT → Grok → Gemini fallback) → streaming
↓ after response
EvaluationAgent → classifies misconceptions (Haiku, ~$0.001)
ContentAgent → pre-generates follow-up exercise
SupervisorAgent → moderates (background, ~$0.001)
Course content is ingested via per-tenant RAG — the tutor cites the institution's own material, not generic content. Details in AI Tutor Overview.
Beyond the tutor's multi-component pipeline, there are separate on-demand administrative agents with their own tools: CourseReviewAgent, AI Course Editor, and Gamification Agent. Only the components that call an LLM use the per-task model config (TenantTaskModelConfig) — Orchestrator/Retrieval/Pedagogical are pure TypeScript (deterministic).
3. Educational CMS with 8 Lesson Types + 6 Interactive Subtypes
Lesson types: slides, video, PDF, rich_text, quiz, assignment, live_class, external_link.
Interactive subtypes: interactive_video (with checkpoints), drag_drop, fill_blanks, flashcard_set (with SM-2 SRS algorithm), timeline, branching_scenario (decision tree).
Auto-course generation: upload a PDF/DOCX/PPTX, an LLM pipeline extracts topics and generates modules with structured lessons. Details in CMS.
4. Quiz Engine with Psychometric Analytics
Cronbach's alpha (internal consistency), point-biserial discrimination, item difficulty, MC/TF distractor analysis, score distribution. Academic integrity: detects tab switches and time anomalies, displays flags in the gradebook. Supports a reusable question bank with 8+ question types.
5. Multi-Provider Live Classes
BigBlueButton (default, iframe embed), Zoom (Server-to-Server OAuth), Microsoft Teams (Graph API), Google Meet (OAuth + Calendar). Unified adapter pattern. Lifecycle webhooks + recording sync cron (fallback). Credentials stored with AES-256-GCM.
6. Full Native Gamification
XP, levels (hierarchical tenant→course config), 11 badge criteria with boolean AND/OR logic, OB 3.0 badges (Open Badges 3.0 — W3C Verifiable Credential), Redis ZSET leaderboards (course/class scope, weekly/monthly/alltime windows), contests with 7 metrics, rewards shop with virtual currency, and a dedicated AI agent for admins to design gamification strategy via chat.
7. Integrated Communication
Forums (per class or course-wide), announcements (with class scoping and auto-notifications), private messages (DMs), real-time in-app notifications, attachments in any channel (forum/announcement/message), institutional email with Handlebars templates in 4 languages, 3 providers (SMTP/Resend/SendGrid).
8. Enterprise SSO + SCIM 2.0
SAML 2.0 (via node-saml), OIDC with PKCE, SCIM 2.0 (14 RFC 7644 endpoints), JIT provisioning, group→course mapping (automatic enrollment), SCIM audit log. Enterprise plan.
9. LTI 1.3 (Tool Provider)
Studeia can be registered as an external tool in Canvas, Blackboard, Moodle, or any LTI 1.3-compliant LMS. OIDC launch, JWT/JWKS, AGS grade passback (grades sync back automatically). Details in LTI integrations.
10. Automations (Built-in Mini-n8n)
Trigger → Condition → Action engine. 3 trigger types: platform event, scheduled cron, and public webhook. 21 supported events, including assignment_submitted, live_class_scheduled, badge_earned, level_up, and contest_ended. 9 actions, including send_email. SSRF prevention, rate limiting, chain depth limit of 5. Public webhook authenticated by token. Details in Automations.
11. B2B Mobile App (React Native/Expo)
Android app published. ~47 /api/mobile/* endpoints with simplified shape and Bearer auth. Dynamic white-label post-login. Push notifications via Expo. SSO via deep linking (studeia://auth/callback). OTA updates via expo-updates.
12. Native LGPD Compliance
GET /api/user/data-export (Art. 18 IV, rate limit 1/24h). DELETE /api/user/account (Art. 18 VI, anonymizes PII, retains learning data for 5 years per fiscal retention rules). Public Privacy/Terms pages in 4 languages. DPA available for Enterprise.
13. Public API with 166 Documented Methods
Dual auth (session + Bearer API key). Granular scopes (courses:read, users:write, etc. — 35 types). Redis rate limiting (3 tiers). Docs at /api/overview.
14. Real-Time AI Tutor Talking Avatar
Opt-in per course (HeyGen LiveAvatar / D-ID, BYO key). See Avatar & TTS.
15. Adaptive Learning
Bayesian ConceptMastery, 3-state misconception lifecycle, spaced repetition. See Adaptive Learning.
16. Multi-Provider AI
Switch Claude/GPT/Grok/Gemini per task, no vendor lock-in. See Multi-provider AI.
Pricing Model
5 B2B plans:
| Plan | Price/mo | Students | Best For |
|---|---|---|---|
| Demo | Free | 1 | Evaluating the platform (Haiku only, 10 msgs/day) |
| Mini | R$250 | 10 | Private tutor, small school |
| Growth | R$1,000 | 50 | Small test-prep course, corporate training |
| Scale | R$1,500 | 100 | Mid-size test-prep course, school |
| Enterprise | Custom quote | Custom (maxStudentsOverride) | University, large corporate L&D |
Additional AI cost: typically R$2–8/student/mo depending on usage (tracked via AiUsageLog per tenant). Admins can switch providers (Claude → GPT → Grok → Gemini) without breaking any feature.
Tech Stack Summary
- Runtime: Next.js 15 App Router, React Server Components, TypeScript strict mode
- Database: PostgreSQL (Supabase) + pgvector (RAG, 1024 dims)
- ORM: Prisma 7 with adapter-pg
- Cache/Queue: Redis 7 + BullMQ
- LLM: Claude (primary) → OpenAI → xAI Grok → Google Gemini (per-tier fallback chain)
- Embeddings: Voyage AI (primary, 1024 dims) → OpenAI (fallback)
- Auth: Supabase Auth with SSR
- Storage: Supabase Storage (public for media, private for submissions)
- Email: SMTP/Resend/SendGrid (factory pattern)
- Live video: BBB/Zoom/Teams/Google Meet (adapter pattern)
- Billing: Stripe + Asaas (PIX)
- Mobile: Expo SDK 52 + NativeWind + Supabase + TanStack Query
- Deploy: Docker + Caddy (on-demand TLS) + Let's Encrypt on a self-managed VPS
- Observability: Sentry (10% session replay, 100% on error)
Compliance and Security
- Native LGPD compliance (data export, account deletion, Enterprise DPA)
- WCAG 2.1 AA (skip-to-content, focus-visible, high-contrast, A11y toolbar, keyboard nav)
- Open Badges 3.0 (W3C Verifiable Credentials)
- LTI 1.3 (1EdTech certified-compatible)
- SCIM 2.0 (RFC 7644)
- AES-256-GCM for OAuth credentials + API keys
- HMAC-SHA256 for OAuth state + impersonation cookies
- Redis rate limiting on all LLM surfaces
- 100 unit/integration tests passing (45 core + 55 web)
Full Feature Map
| Area | Pages |
|---|---|
| Platform | Multi-tenancy · White-label · Multi-provider AI · Mobile app · Security & LGPD · Accessibility · i18n · Enterprise SSO |
| AI Tutor | Pipeline · RAG ingestion · Adaptive learning · Avatar & TTS · Safety supervisor |
| CMS | Courses & lessons · Lesson types · Interactive content · AI course builder · Media library |
| Assessment | Quiz engine · Question bank · Gradebook |
| Engagement | Gamification · Live classes · Communication · Email |
| Operations | Automations · Reports & risk · API · LTI 1.3 |
Known Limitations
In the spirit of transparency about trade-offs:
- iOS app: still on the roadmap (requires an Apple Developer Program account at $99/yr)
- Voice input (STT): speaking to the tutor (speech → text) is already implemented in B2B (dictation); the talking avatar is output-only, and full two-way voice conversation remains on the roadmap
- AI-generated pre-recorded video lessons (Synthesia): not implemented
- Cryptographic badge signing & native LinkedIn share button: roadmap (badges are verifiable via public URL/API today)
- Third-party plugins: Studeia does not have a plugin ecosystem like Moodle. Customization is done via the public API.
- On-premise: Studeia is SaaS-only. We do not yet serve regulations that require on-premise hosting.
None of these gaps affect the core product — AI tutor with per-course RAG, full CMS, psychometric assessment, gamification, enterprise SSO, and native LGPD compliance are all in production today.
Start with a Trial, Not a Sales Call
The fastest way to evaluate Studeia is not to read this page — it's to create a free demo account (2 minutes, no credit card required), upload real content, and ask the AI tutor your first question. Everything described above is immediately testable, and pricing is public.