RoboLearn: AI-Native Textbook Platform 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.
┌─────────────────────────────────────────────────────────────┐
│ 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)
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
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
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)
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
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
Feature
Description
Book isolation
Separate knowledge folders per book
Shared infrastructure
Common auth, RAG, components
Second book
"CoLearning Python: The AI-Driven Way"
Feature
Description
White-label
Custom branding per institution
Bulk licensing
Student seat management
LMS integration
Grade passback, SSO
Feature
Description
Mobile app
React Native
Offline mode
Downloaded content
Real ROS 2
Beyond MockROS for advanced users
Marketplace
Third-party authors
Tier
Price
Features
Free
$0
Read content, basic exercises
Professional
$49/month
RAG chat, personalization, multilingual, certificates
Annual
$399/year
Professional features + 30% savings
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
Model
Split
Free Books
0% platform fee
Paid Books
70% author / 30% platform
Enterprise Licensing
60% author / 40% platform (includes support overhead)
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
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
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
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.