Your team makes promises in Slack and Gmail every day. Most of them disappear.
Sentinel automatically extracts every delivery commitment, scores its risk of slipping, and alerts the right owner — before the deadline is missed.
Watch Demo · Quick Start · Features · Tech Stack · Architecture
Watch Sentinel in action — commitment extraction, risk scoring, and smart alerts.
| Feature | Description |
|---|---|
| 🔍 Passive Commitment Extraction | Silently reads Slack channels and Gmail threads — extracts every promise with owner, due date, and confidence score. Zero input from the team. |
| 📊 4-Factor Risk Scoring | Every commitment scored 0–100 based on deadline proximity, dependency chain, owner silence, and current workload. Updated every 6 hours. |
| 🔔 Smart Owner Alerts | When risk crosses threshold, the owner — not the manager — receives a Slack DM with two options: "I'm On It" or "Draft Update" (AI writes it). |
| 🕸 Dependency Graph | Full-page interactive network visualization. Nodes = team members. Edges = commitments. Red = at risk. Click any node for full commitment list. |
| 🌅 Morning Brief | Daily situational awareness digest — surfaces critical updates, risks, and blocked teammates before the workday starts. |
| 🧊 Glass Command Dashboard | Ultra-premium interface with real-time Execution Health Score (0–100) for leadership — one glance, full picture. |
| 🔗 Multi-Platform Integrations | Slack · Gmail · GitHub · Jira · Notion · Linear · Google Calendar · Microsoft Teams |
| Package | Version | Purpose |
|---|---|---|
| React | 19 | UI Framework |
| Vite | 6 | Build Tool |
| TypeScript | 5.8 | Type Safety |
| Tailwind CSS | 4 | Styling |
| Framer Motion | 12 | Animations |
| TanStack Query | 5 | Data Fetching & Caching |
| Package | Version | Purpose |
|---|---|---|
| FastAPI | 0.115 | REST API Framework |
| Python | 3.11 | Runtime |
| Uvicorn | Latest | ASGI Server |
| Supabase | 3 | Database + Auth + RLS |
| Groq API | Latest | LLM Inference (Llama 3.3 70B) |
| Google Gemini | 1.5 | Fallback LLM |
| Celery | Latest | Background Task Queue |
| Upstash Redis | Latest | Message Broker |
| python-jose | Latest | JWT Authentication |
| Service | Purpose |
|---|---|
| Vercel | Frontend hosting · Global CDN |
| Render | Backend hosting · Async workers |
| Supabase Cloud | Managed PostgreSQL · Row-Level Security |
| Upstash | Serverless Redis · Task queue |
| UptimeRobot | Uptime monitoring · Keep-alive |
| 🔥 IBM Cloud AMD MI300X | Enterprise LLM inference (Roadmap) |
┌─────────────────────────────────────────────────────────┐
│ CLIENT LAYER (React 19 / Vite / TS) │
│ Dashboard · Commitments · Graph · Integrations · Auth │
└──────────────────────────┬──────────────────────────────┘
│ HTTPS / REST API
┌──────────────────────────▼──────────────────────────────┐
│ API & AUTH LAYER (FastAPI + JWT) │
│ /auth /dashboard /commitments /webhooks /cron │
│ OAuth 2.0: Slack · Gmail · GitHub │
└──────────┬───────────────────────────┬───────────────────┘
│ │
┌──────────▼──────────┐ ┌───────────▼────────────────────┐
│ DATA LAYER │ │ AI PROCESSING ENGINE │
│ Supabase │ │ Groq LPU · Llama 3.3 70B │
│ PostgreSQL + RLS │ │ Risk Engine · LLM Client │
│ 5 core tables │ │ → AMD Instinct MI300X roadmap │
└─────────────────────┘ └────────────────┬────────────────┘
│
┌──────────────────────────────────────────▼────────────────┐
│ EXTERNAL INTEGRATIONS │
│ Slack API · Gmail API · GitHub OAuth · Jira │
└────────────────────────────────────────────────────────────┘
Slack Message / Gmail
│
▼
Webhook / Polling Capture ── FastAPI endpoint
│
▼
Preprocessing ── De-duplicate · Enrich metadata
│
▼
Groq LPU (Llama 3.3 70B) ── Returns structured JSON:
│ { commitment, owner_email,
│ due_date, confidence,
│ source_channel }
▼
Risk Engine ── Score 0–100
│ (deadline · dependencies · silence · workload)
▼
Supabase ── Store with RLS
│
▼
Dashboard + Alerts ── Owner notified if score > threshold
| Requirement | Version |
|---|---|
| Node.js | v18+ |
| Python | 3.10+ |
| npm or yarn | Latest |
| pip | Latest |
git clone https://github.com/your-org/sentinel.git
cd sentinel# Navigate to frontend
cd SENTINEL-Frontend
# Install dependencies
npm install
# Configure environment
cp .env.example .env
# Add your API keys to .env
# Start development server
npm run devFrontend runs at
http://localhost:3000
# Navigate to backend
cd SENTINEL-Backend
# Create and activate virtual environment
python -m venv venv
# Windows
venv\Scripts\activate
# Mac / Linux
source venv/bin/activate
# Install dependencies
pip install -r requirements.txt
# Configure environment
cp .env.example .env
# Add your credentials to .env
# Start development server
uvicorn app.main:app --reloadBackend API runs at
http://localhost:8000Interactive API docs athttp://localhost:8000/docs
| Variable | Description | Required |
|---|---|---|
VITE_API_URL |
Backend API base URL | ✅ |
VITE_GEMINI_API_KEY |
Gemini API key for AI features | ✅ |
| Variable | Description | Required |
|---|---|---|
SUPABASE_URL |
Supabase project URL | ✅ |
SUPABASE_KEY |
Supabase service role key | ✅ |
GEMINI_API_KEY |
Google Gemini API key | ✅ |
GROQ_API_KEY |
Groq API key (Llama 3.3 70B) | ✅ |
JWT_SECRET |
Secret for JWT token signing | ✅ |
GOOGLE_CLIENT_ID |
Google OAuth client ID | ✅ |
GOOGLE_CLIENT_SECRET |
Google OAuth client secret | ✅ |
GITHUB_CLIENT_ID |
GitHub OAuth client ID | ✅ |
GITHUB_CLIENT_SECRET |
GitHub OAuth client secret | ✅ |
SLACK_CLIENT_ID |
Slack OAuth client ID | ✅ |
SLACK_CLIENT_SECRET |
Slack OAuth client secret | ✅ |
REDIS_URL |
Upstash Redis connection URL | ✅ |
SENTINEL/
├── SENTINEL-Frontend/ # React + Vite frontend
│ ├── src/
│ │ ├── components/ # Reusable UI components
│ │ ├── pages/ # Page-level components
│ │ │ ├── Dashboard.tsx # Execution Health Score
│ │ │ ├── Commitments.tsx # Full commitment list
│ │ │ ├── Graph.tsx # Dependency graph
│ │ │ └── Integrations.tsx # OAuth connection cards
│ │ ├── context/ # React context providers
│ │ ├── hooks/ # Custom React hooks
│ │ └── lib/ # Utilities & helpers
│ └── index.html
│
├── SENTINEL-Backend/ # FastAPI backend
│ ├── app/
│ │ ├── api/ # API route handlers
│ │ │ ├── auth.py # JWT auth endpoints
│ │ │ ├── dashboard.py # Metrics aggregation
│ │ │ ├── commitments.py # Commitment CRUD
│ │ │ ├── webhooks.py # Slack event ingestion
│ │ │ └── cron.py # Risk re-scoring jobs
│ │ ├── services/ # Business logic
│ │ │ ├── extractor.py # LLM commitment extraction
│ │ │ └── risk_engine.py # Risk scoring algorithm
│ │ ├── models/ # Pydantic data models
│ │ ├── integrations/ # External API clients
│ │ │ ├── slack.py
│ │ │ ├── gmail.py
│ │ │ └── github.py
│ │ └── db/ # Supabase utilities
│
├── docs/
│ └── assets/ # Screenshots & demo video
│ ├── SENTINEL.mp4
│ ├── Dashboard.png
│ ├── Commitment Page.png
│ ├── Dependency Graph Page.png
│ └── Integration Page.png
│
└── README.md
Sentinel follows the Zenith of Beauty principle — where form meets function with absolute clarity.
- Every interaction is designed to feel physical and immediate
- Every blur is mathematically balanced for readability
- Every animation communicates system status, not decoration
- The Glass Command aesthetic provides depth without distraction
| Phase | Status | Description |
|---|---|---|
| Prototype · Free Tiers | ✅ Live | Groq · Supabase · Render · Vercel |
| Gmail OAuth Polling | ✅ Live | gmail.readonly · httpx polling |
| Slack Event Webhooks | ✅ Live | Real-time message ingestion |
| 4-Factor Risk Engine | ✅ Live | Deterministic 0–100 scorer |
| AMD EPYC™ Cloud Deploy | 🔄 Phase 2 | Dedicated VM · 4 vCPU · 8GB |
| AMD Instinct™ MI300X | 🔜 Phase 3 | Self-hosted Llama 3.3 70B via IBM Cloud |
| ROCm™ Inference Stack | 🔜 Phase 3 | VPC-native · Zero data leaves infra |
| Multilingual Extraction | 🔜 Phase 2 | Hindi · Spanish · French · Arabic |
| Microsoft Teams Integration | 🔜 Phase 2 | OAuth + message polling |
| Phase | Monthly Cost | Notes |
|---|---|---|
| 🟢 Prototype | ₹0 | All free tiers |
| 🟡 Growth (50 seats) | ₹21,385 | Profitable at 50 workspaces × ₹499 |
| 🔴 Enterprise (AMD MI300X) | ₹728–₹1,092/hr | 4–6× cheaper than cloud API at scale |
Exchange rate: $1 = ₹91
This project is licensed under the MIT License — see the LICENSE file for details.
Built with ❤️ by Quantumstacks Lab · AMD Slingshot Hackathon 2026
🐛 Report Bug ·
💬 Discussions ·
⭐ Star this repo
Powered by AMD EPYC™ · Roadmap: AMD Instinct™ MI300X + ROCm™ · IBM Cloud




