Skip to content

Book + RAG ChatAgent + SSO + Compunding RI Intelligence built in 2 days. Robolearn is what AI-driven development with Spec-Driven methodology and Reusable Intelligence makes possible.

Notifications You must be signed in to change notification settings

mjunaidca/robolearn

Repository files navigation

RoboLearn: AI-Native Textbook Platform

The Thesis

The software industry has disrupted itself. Spec-Driven Development with Reusable Intelligence (SDD-RI) transforms what traditionally takes months into days—not through faster coding, but through compounding intelligence.

This hackathon isn't just about winning 300 points. It's about launching a platform.


What We're Actually Building

Traditional Timeline Our Timeline
Book content: 6-18 months 48 hours
Author platform: 3-6 months Week 1-2
Multi-book infrastructure: 6-12 months Month 1
Institutional features: 12+ months Month 2

Why? Because every hour invested in reusable intelligence compounds. The lesson-writer agent that creates Module 1 creates Module 4 at the same speed. The skills that power RoboLearn power the next ten books.


Platform Vision

┌─────────────────────────────────────────────────────────────┐
│                    RoboLearn Platform                        │
├───────────────────┬───────────────────┬─────────────────────┤
│     STUDENTS      │      AUTHORS      │    INSTITUTIONS     │
│                   │                   │                     │
│ Personalized      │ AI-assisted       │ White-label         │
│ Hardware-aware    │ Days not months   │ Analytics           │
│ Multilingual      │ Revenue share     │ Curriculum control  │
│ Interactive       │ Agent workforce   │ Bulk licensing      │
└───────────────────┴───────────────────┴─────────────────────┘

Hackathon Deliverables (Sunday 6 PM)

Scoring Target: 300/300

Requirement Points Deliverable
Book + RAG Chatbot 100 4 modules, context-aware chat
Reusable Intelligence 50 Skills, agents, knowledge, MCP configs
Auth + Onboarding 50 Better-Auth, hardware survey, profile-based filtering
Personalization 50 AI rewrites content for user context
Urdu Translation 50 Toggle between English/Urdu
Total 300

Student Experience

Signup → Hardware Survey → Personalized Content
                ↓
    ┌────────┬────────┬────────────┐
    │ Learn  │Visualize│Personalize │  ← 3-Tab UI
    │ (MDX)  │(Diagram)│   (AI)     │
    └────────┴────────┴────────────┘
                ↓
    ┌─────────────────────────────┐
    │    Interactive Python Lab    │
    │  Pyodide + MockROS + Robot   │
    └─────────────────────────────┘
                ↓
    ┌─────────────────────────────┐
    │      RAG Chat Sidebar        │
    │  Context-aware • Select-ask  │
    └─────────────────────────────┘
                ↓
           🔄 EN ↔ UR

Technical Architecture

Stack

Layer Choice Why
Frontend Docusaurus MDX-native, fast builds
Hosting GitHub Pages → Cloudflare Free, global CDN
Backend FastAPI + Cloud Run Serverless, scales to zero
Database Neon Postgres Profiles, hardware configs
Vector DB Qdrant Cloud RAG embeddings
Auth Better-Auth Modern, official MCP server
AI OpenAI Agents SDK Chat, personalization

Reusable Intelligence Structure

.claude/
├── skills/                           # HOW (reusable patterns)
│   ├── authoring/                    # Content creation skills
│   │   ├── lesson-generator/         # Generate lessons (4-layer framework)
│   │   ├── assessment-builder/       # Create quizzes and assessments
│   │   ├── learning-objectives/      # Design learning objectives
│   │   ├── mermaid-diagram/          # Generate educational diagrams
│   │   ├── urdu-translator/          # Translate content to Urdu
│   │   ├── quiz-generator/           # Generate quiz questions
│   │   ├── summary-generator/        # Create lesson summaries
│   │   ├── notebooklm-slides/        # Generate slide decks
│   │   └── concept-scaffolding/      # Scaffold complex concepts
│   │   └── visual-asset-workflow/    # Create educational visuals
│   └── engineering/                  # Platform development skills
│       ├── pyodide-exercise/         # Browser-based Python exercises
│       ├── docusaurus-deployer/      # Deploy to GitHub Pages
│       ├── frontend-design/          # UI component design
│       ├── image-generator/          # Generate images with Gemini
│       ├── mcp-builder/              # Create MCP servers
│       ├── skill-creator/            # Create new skills
│       ├── chatkit-integration/       # ChatKit framework integration patterns
│       └── session-intelligence-harvester/  # Capture learnings
│
├── agents/                           # WHO (autonomous workers)
│   ├── super-orchestra.md            # General workflow orchestration
│   ├── authoring/                    # Content creation agents
│   │   ├── content-implementer.md    # Implement lessons from specs
│   │   ├── chapter-planner.md        # Plan chapter structure
│   │   ├── educational-validator.md  # Validate constitutional compliance
│   │   └── validation-auditor.md     # Quality validation before publish
│   └── engineering/                  # Platform development agents
│       ├── spec-architect.md         # Design specifications
│       └── chatkit-integration-agent.md  # ChatKit integration workflow
│
├── commands/                         # Slash commands (/sp.*)
└── .mcp.json                         # MCP server configuration

# Domain knowledge lives in authoritative sources:
# - requirement.md (course structure, hardware specs)
# - .specify/memory/constitution.md (principles, tiers)
# - README.md (platform vision)

MCP Strategy

Server Use Rationale
Better-Auth MCP Auth implementation Active introspection — generates schemas, supersedes docs
Context7 Library docs Generalist for React, FastAPI, Pyodide
Tavily Research Synthesized answers for content generation
DeepWiki Repo understanding Understand panaversity base template

Execution Plan (10 Hours)

Phase 1: Foundation + Intelligence (Hour 0-2) ✅ COMPLETE

Task Deliverable Status
1.1 Fork repo, rename to robolearn, verify build
1.2 Create folder structure (skills, agents, knowledge, mcp)
1.3 Write knowledge files (vocabulary, hardware-tiers, course-structure)
1.4 Write skill files (lesson-generator, hardware-filter, urdu-translator)
1.5 Write agent files (lesson-writer, rag-builder)
1.6 Configure MCP servers
1.7 Content cleanup, rebrand, navigation
1.8 Component stubs, first deploy
1.9 Homepage redesign (Industrial Confidence design system)

Exit: ✅ Live at username.github.io/robolearn with intelligence infrastructure + redesigned homepage

Phase 2: Content Generation (Hour 2-4) ✅ COMPLETE

Task Deliverable Status
2.1 Module 1: ROS 2 Foundations (7 chapters, 25 lessons)
2.2 Module 2: Gazebo/Unity Simulation (6 chapters, 22 lessons)
2.3 Module READMEs and chapter structure
2.4 8 authoring skills created

Exit: ✅ 47 lessons across 2 modules with complete skill infrastructure

Extensions (moved to Phase 5):

  • Create Mermaid/React Flow diagrams
  • Add hardware-filtered sections

Phase 3: Auth + Profiles (Hour 4-5)

Task Deliverable
3.1 Better-Auth setup (use official MCP)
3.2 Neon Postgres schema
3.3 Survey component

Exit: Users can signup, complete survey, see filtered content

Phase 4: Backend + RAG (Hour 5-7) ✅ COMPLETE

Task Deliverable Status
4.1 FastAPI app structure
4.2 Qdrant collection setup
4.3 Embedding pipeline (content → vectors)
4.4 OpenAI Agents SDK config
4.5 Deploy to Cloud Run

Exit: ✅ RAG chatbot answering questions with book context + visual enhancements

Extensions Completed:

  • ChatKit server integration with PostgreSQL persistence
  • Context injection (user profile, page context, conversation history)
  • Streaming responses for real-time UX
  • Complete specifications reverse-engineered (specs/007-chatkit-server/)

Phase 5: Chat UI (Hour 7-8) ✅ COMPLETE

Task Deliverable Status
5.1 ChatKit widget component
5.2 Current page context injection
5.3 Select-to-ask functionality
5.4 Hardware-aware responses
5.5 User authentication integration
5.6 Personalization menu
5.7 Script loading detection

Exit: ✅ Functional ChatKit widget with context awareness, text selection "Ask", and user personalization

Extensions Completed:

  • ChatKit React component integration
  • Custom fetch interceptor for auth and metadata
  • Page context extraction (URL, title, headings, meta tags)
  • User profile context transmission
  • Complete specifications reverse-engineered (specs/008-chatkit-ui-widget/)
  • Reusable intelligence harvested (skill + agent)

Phase 6: Bonus Features (Hour 9-9.5) ✅ COMPLETE

Task Deliverable Status
6.1 Urdu translation or Multi Language
6.2 LanguageToggle component

Exit: ✅ Full 300-point feature set

Completed Features:

  • LanguageToggle component with locale switching (/en/ and /ur/ routes)
  • Docusaurus i18n plugin for auto-translation (Gemini API)
  • RTL (Right-to-Left) CSS support for Urdu content
  • Language preference persistence (localStorage)
  • Complete specification (specs/006-i18n-auto-translate-gemini/)

Phase 7: Ship (Hour 9.5-10) ✅ MOSTLY COMPLETE

Task Deliverable Status
7.1 End-to-end testing
7.2 90-second demo video 🟡 Pending
7.3 README with setup instructions
7.4 Submit 🟡 Pending

Exit: ✅ Hackathon submission ready (demo video pending)

Completed:

  • End-to-end testing completed for ChatKit integration
  • Comprehensive README with setup instructions
  • Complete documentation (specs, ADRs, PHRs)
  • Reusable intelligence infrastructure

Phase 8: Interactive Lab (Hour 8-9)

Task Deliverable
6.1 PythonRunner component (Pyodide)
6.2 MockROSBridge class
6.3 RobotViewer component
6.4 Wire up: code → mock ROS → robot responds
6.5 Personalization endpoint
6.6 Personalize tab in content

Exit: Students write ROS-like code, see robot react


Post-Hackathon Roadmap

Week 1-2: Author Platform

Feature Description
Author Dashboard Book management, chapter organization
Agent Studio Configure lesson-writer, review AI drafts
Analytics Reader engagement, chat queries, hardware distribution

Month 1: Multi-Book

Feature Description
Book isolation Separate knowledge folders per book
Shared infrastructure Common auth, RAG, components
Second book "CoLearning Python: The AI-Driven Way"

Month 2: Institutional

Feature Description
White-label Custom branding per institution
Bulk licensing Student seat management
LMS integration Grade passback, SSO

Month 3+: Scale

Feature Description
Mobile app React Native
Offline mode Downloaded content
Real ROS 2 Beyond MockROS for advanced users
Marketplace Third-party authors

Revenue Model

Individual Learners

Tier Price Features
Free $0 Read content, basic exercises
Professional $49/month RAG chat, personalization, multilingual, certificates
Annual $399/year Professional features + 30% savings

Corporate Training

Tier Price Features
Team $199/month 10 seats, progress tracking, admin dashboard
Department $799/month 50 seats, custom learning paths, analytics
Enterprise Custom Unlimited seats, SSO, dedicated support, SLA

Institutional (Universities, Bootcamps)

Tier Price Features
Academic $2,500/year 200 students, LMS integration, grade passback
Campus $10,000/year 1,000 students, white-label option, curriculum control
Enterprise $50,000+/year Unlimited, custom development, dedicated success manager

Author Revenue

Model Split
Free Books 0% platform fee
Paid Books 70% author / 30% platform
Enterprise Licensing 60% author / 40% platform (includes support overhead)

Competitive Moat

Why this is hard to copy:

Layer Moat
Skills Compound with every book authored
Knowledge Domain expertise encoded
Agents Workflows refined through use
Network Zia Khan's 50K+ student distribution
Data Chat queries reveal what students struggle with

Every book makes the platform smarter. Every student interaction improves the RAG. The intelligence compounds.


Risk Mitigation

Risk Mitigation
Deadline pressure Phases ordered by point value, cut from bottom
RAG quality Start basic, iterate
Live demo fails Pre-recorded backup video
Content thin 3 polished lessons > 10 rough ones
MockROS feels fake Frame as "pedagogical simulation" — judges evaluate concept

Success Metrics

Phase Metric Target
Hackathon Score 300 points
Week 1 Author dashboard MVP live
Month 1 Books 2 live
Month 1 Students 500 active
Month 3 MRR $2,000
Month 6 Students 10,000

The Bottom Line

We don't just submit a hackathon project. We launch a platform.

The same intelligence that builds RoboLearn builds the next ten books. The same agents that write Physical AI lessons write Python lessons. The same infrastructure that serves 100 students serves 100,000.

This is what AI-driven development with Spec-Driven methodology and Reusable Intelligence makes possible.


About

Book + RAG ChatAgent + SSO + Compunding RI Intelligence built in 2 days. Robolearn is what AI-driven development with Spec-Driven methodology and Reusable Intelligence makes possible.

Topics

Resources

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published