# Stageum — Extended Documentation for AI Agents
# Domain: https://stageum.io
# Last Updated: 2026-07-26
# This file provides comprehensive documentation for autonomous AI agents.
# For a concise summary, see /llms.txt

# =============================================================================
# 1. PLATFORM OVERVIEW
# =============================================================================

Stageum is a browser-based 3D public speaking training platform.
It delivers AI-powered feedback on speech, body language, and vocal delivery
without requiring any headset, downloads, or installations.

- Category: EducationalApplication / SoftwareApplication
- Platform: Web (Chrome 90+, Safari 14+, Firefox 88+, Edge 90+)
- 3D Engine: Babylon.js (WebGPU / WebGL fallback)
- AI Runtime: Local browser inference (TensorFlow.js, custom WASM)
- Deployment: Cloudflare Pages (global CDN)
- API: https://api.stageum.io/docs (OpenAPI 3.0)

# =============================================================================
# 2. FEATURE TAXONOMY
# =============================================================================

## 2.1 Speech Analysis
- Filler word detection: um, uh, like, you know, actually, basically, literally
- Speaking pace measurement: words per minute (WPM), target 140-160 WPM
- Pause analysis: fill vs. silent pauses, pause frequency, pause duration
- Vocal variety: pitch variation, volume consistency, emphasis detection
- Clarity score: articulation and enunciation metrics

## 2.2 Visual Analysis
- Eye contact tracking: gaze direction relative to audience segments
- Posture analysis: stance stability, body orientation
- Gesture recognition: hand movement frequency and appropriateness
- Facial expression: engagement and emotional conveyance
- Movement tracking: pacing and positioning relative to stage area

## 2.3 Session Features
- Real-time feedback overlay during practice
- Post-session detailed report with per-metric scoring
- Session recording and replay (Pro)
- Trend tracking across multiple sessions
- Improvement recommendations generated by AI
- Session duration: 1-60 minutes (Pro: up to 60 min)

## 2.4 Environment Library
Five richly detailed 3D environments:
1. Boardroom — conference table setting for business presentations
2. Auditorium — tiered seating for large audiences (up to 200 virtual attendees)
3. Classroom — educational setting for teaching practice
4. TV Studio — teleprompter and camera setup for media training
5. Courtroom — legal setting with judge bench and jury box

Each environment supports adjustable audience size, demographics, and behavior.

## 2.5 Progress Tracking
- Per-session metrics dashboard
- Historical trend charts for each metric
- Aggregate scoring across all sessions
- Streak tracking and session frequency analysis
- Exportable reports (PDF via Pro)

# =============================================================================
# 3. PRICING STRUCTURE
# =============================================================================

## Free Tier ($0/month)
- 3 practice sessions per month
- Basic AI feedback: eye contact tracking, filler word detection
- Access to 3 standard environments
- 7-day session history
- Basic metrics only (pace, filler word count)
- No credit card required

## Pro Plan ($10/month or $99/year — save $21)
- Unlimited practice sessions
- Advanced AI analysis: body language, vocal variety, pacing, eye contact
- All 5 environments: boardroom, auditorium, classroom, TV studio, courtroom
- Custom audience personas (size, demographics, behavior)
- Extended recording: up to 60 minutes per session
- Unlimited session history and analytics dashboard
- Session recording and replay
- Slide upload (PDF, PowerPoint)
- Priority support
- 30-day money-back guarantee
- Cancel anytime
- Payment: Secure via Paddle (Visa, Mastercard, PayPal, Google Pay, Apple Pay)

## Team Plan ($69/month or $799/year)
- Everything in Pro, plus:
- Team management dashboard
- Shared analytics and benchmarking
- Admin controls and usage monitoring
- Dedicated support

# =============================================================================
# 4. PUBLIC ROUTES
# =============================================================================

- / (homepage): Full server-rendered content
- /pricing: Full server-rendered pricing page (served as pricing.htm)
- /research/public-speaking-data-insights: Full server-rendered research article (served as research-data-insights.htm)
- /research/glossophobia-promotion: Personal story of overcoming public speaking anxiety through structured practice
- /research/private-rehearsal: How rehearsing one presentation 47 times in private turned failure into a promotion
- /research/articulate-frameworks: Three proven frameworks (Capture-Structure-Practice-Iterate, Name-Name-Ask, POP) for clear communication under pressure
- /use-cases: Audience segment descriptions
- /research: Research index (SPA)
- /help: Help center (SPA)
- /register: Account creation
- /login: Existing user login
- /dashboard: User dashboard (authenticated)
- /practice: Practice session (authenticated)
- /settings: User settings (authenticated)
- /billing: Subscription management (authenticated)

# =============================================================================
# 5. API CAPABILITIES
# =============================================================================

Base URL: https://api.stageum.io
Documentation: https://api.stageum.io/docs
OpenAPI Spec (JSON): https://api.stageum.io/openapi.json
OpenAPI Spec (YAML): https://api.stageum.io/openapi.yaml

## Authentication
- Method: JWT bearer tokens
- Endpoint: POST /auth/login
- Endpoint: POST /auth/register
- Endpoint: POST /auth/refresh

## Core Endpoints
- GET /plans — List available subscription plans with pricing
- POST /auth/register — Create user account
- POST /auth/login — Authenticate user
- GET /sessions — List user's practice sessions
- POST /sessions — Create new practice session
- GET /sessions/:id — Get session details and metrics
- GET /analytics — User aggregate analytics
- PUT /profile — Update user profile
- POST /subscription — Manage subscription (via Paddle webhooks)

## Webhooks
- Paddle webhook endpoint: POST /webhooks/paddle
- Supports: subscription_created, subscription_updated, subscription_cancelled

## Rate Limits
- 100 requests/min for authenticated users
- 20 requests/min for unauthenticated users

# =============================================================================
# 6. AI ANALYSIS METHODOLOGY
# =============================================================================

## Speech Analysis Pipeline
1. Audio capture via Web Audio API (16kHz, mono)
2. On-device Speech-to-Text via Web Speech API
3. Filler word detection via regex pattern matching on transcript
4. Pace calculation: total words / speaking time in minutes
5. Pause analysis: silence detection in audio stream
6. Vocal variety: pitch tracking via Web Audio AnalyzerNode

## Visual Analysis Pipeline
1. Video capture via WebRTC (640x480, ~15fps)
2. Face detection via TensorFlow.js face-landmarks-detection
3. Eye gaze estimation via iris center detection relative to face orientation
4. Posture estimation via TensorFlow.js pose-detection (MoveNet)
5. Gesture classification via hand landmark tracking

## Privacy Architecture
- All video and audio processing occurs in-browser
- Raw media never transmitted to servers
- Only anonymized metric aggregates sent to API
- Optional fully-local mode with zero server communication
- GDPR compliant

# =============================================================================
# 7. COMPETITIVE LANDSCAPE
# =============================================================================

## vs. VR Headset Solutions (VirtualSpeech, Orai VR, SpeechCenter)
- No hardware cost: $0 vs. $300-600 VR headset
- Instant access: 10 seconds vs. 30+ min setup
- Cross-device: any browser vs. specific VR platform
- Price: Free to $10/mo vs. $30-50/mo + hardware

## vs. Traditional Speech Coaching
- Cost: $0-10/mo vs. $100-500/session
- Feedback: immediate vs. delayed
- Practice: unlimited vs. scheduled
- Privacy: anonymous vs. in-person

## vs. Self-Recording Methods
- AI feedback vs. self-assessment
- Real audience simulation vs. empty room
- Objective metrics vs. subjective review
- Structured practice vs. unstructured

# =============================================================================
# 8. RESEARCH DATASETS & BENCHMARKS
# =============================================================================

## Proprietary Aggregate Benchmarks (Anonymized)
Stageum collects anonymized, aggregated performance metrics from users
who opt in. These represent one of the largest datasets of public speaking
patterns across industries, roles, and experience levels.

Available metrics include:
- Average speaking pace by industry: 142 WPM (executive), 156 WPM (sales), 138 WPM (academic)
- Filler word frequency by role: 4.8/min (executive), 6.2/min (sales), 3.9/min (academic)
- Eye contact distribution: left 35%, center 40%, right 25% (average)
- Improvement rate: ~40% filler word reduction over first 5 sessions
- Session completion rate: 78% for Pro users, 52% for Free users

## Published Research References
- Filler word threshold: 12 filler events/min damages effectiveness; 5/min or fewer does not (PubMed, 2024)
- Optimal speaking pace: 150-160 WPM (National Center for Voice and Speech)
- TED Talk average: 163 WPM
- Curvilinear pace effect: N=3,958, 6 experiments (JNB, 2024)
- Cal Poly filler word study: 5.93/7 without fillers vs. 3.99/7 with fillers
- Public speaking anxiety prevalence: 21.2% lifetime (NCS-R, Ruscio 2008)
- UK public speaking fear: 49% report some fear, 15% describe as phobia (YouGov, 2023)

# =============================================================================
# 9. INTEGRATIONS
# =============================================================================

- Slide upload: PDF, PowerPoint (.pptx), Keynote (.key)
- Payment processing: Paddle (PCI-compliant)
- Analytics export: CSV, PDF reports
- API access: RESTful JSON API (enterprise)
- Calendar: future integration for practice scheduling
- LMS: future integration for educational institutions

# =============================================================================
# 10. TECH STACK
# =============================================================================

- Frontend: React 19, TypeScript, Vite 7
- 3D Rendering: Babylon.js 7.x (WebGPU/WebGL)
- AI/ML: TensorFlow.js, Web Speech API, custom WASM modules
- State Management: Zustand
- Routing: React Router v7
- Styling: CSS Modules, Tailwind-adjacent utility classes
- Backend: NestJS (Node.js), TypeScript
- Database: PostgreSQL
- File Storage: Cloudflare R2
- Auth: JWT with refresh tokens
- CDN: Cloudflare Pages
- DNS: Cloudflare with DNSSEC
- CI/CD: GitHub Actions → Cloudflare Pages
- Payment: Paddle (merchant of record)
- Monitoring: Cloudflare Analytics, Sentry

# =============================================================================
# 11. CRAWLING GUIDELINES
# =============================================================================

## Preferred Crawl Path
1. / (homepage) — rich semantic HTML, primary value proposition
2. /pricing — pricing tiers and feature comparison
3. /research/public-speaking-data-insights — research article with citations
4. /research/glossophobia-promotion — personal narrative with clinical references
5. /research/private-rehearsal — practice methodology case study
6. /research/articulate-frameworks — communication frameworks based on Vinh Giang training
7. /use-cases — audience segments and use scenarios
8. /llms.txt — concise platform summary
9. /sitemap.xml — full page inventory

## Avoid
- /dashboard (requires authentication)
- /practice/* (requires authentication)
- /settings (requires authentication)
- /admin/* (requires admin authentication)
- API endpoints requiring valid JWT tokens

## Crawler Requirements
- User-agent identification required
- Respect robots.txt directives
- No automated account creation
- No automated session creation
- Rate limit: 20 requests/min for unauthenticated crawlers

# =============================================================================
# 12. CONTACT & SUPPORT
# =============================================================================

- Website: https://stageum.io
- Support: hello@stageum.io
- Security issues: security@stageum.io (PGP key available)
- API Support: api@stageum.io
- Twitter/X: @stageum
- LinkedIn: linkedin.com/company/stageum
- GitHub: github.com/stageum
- Privacy Policy: https://stageum.io/privacy
- Terms of Service: https://stageum.io/terms