autonode — AI Dead Zone Hunter

Industrial IoT mesh network monitoring with autonomous AI agents built on Jac / Jaseci

---

What It Does

autonode is a real-time IoT sensor monitoring platform that uses autonomous AI agents to detect, diagnose, and respond to sensor failures in an industrial warehouse — automatically, without human intervention.

When a sensor goes dark, a pipeline of 5 specialized agents fires:

Agent	Role
HeartbeatMonitor	Continuously scans all sensors for missed pings
DeadZoneMapper	Maps the geographic spread of signal loss
RootCauseAnalyzer	Diagnoses the failure (battery, obstruction, overload)
ReroutingAgent	Attempts automatic mesh rerouting for self-healing
DispatchAgent	Generates a work order if hardware intervention is needed

---

Architecture

┌─────────────────────────────────┐
│  React Frontend (localhost:3000)│
│  Live SVG warehouse map         │
│  Real-time agent feed           │
│  Work order display             │
└────────────┬────────────────────┘
             │ HTTP (REST)
┌────────────▼────────────────────┐
│  Jac Backend (localhost:8000)   │
│  jac start server.jac           │
│                                 │
│  Graph: Warehouse               │
│    └── Zone (x4)                │
│          └── Router (x4)        │
│                └── Sensor (x24) │
└─────────────────────────────────┘

The backend is built entirely in Jac — a graph-native AI programming language. The warehouse is modeled as a persistent node graph, and each agent is a walker that traverses the graph to detect and respond to anomalies.

---

Tech Stack

• Backend: Jac 0.11.2 / Jaseci stack
• Frontend: React 18, SVG-based live map
• API: Auto-generated REST endpoints via jac start
• Sensor Graph: 24 sensors, 4 routers, 4 zones

---

Setup

Prerequisites

• Python 3.11+
• Node.js 18+
• jaclang and jac-cloud installed

pip install jaclang jac-cloud

Backend

cd backend
python -m venv .venv
.venv\Scripts\activate        # Windows
source .venv/bin/activate     # Mac/Linux

pip install jaclang jac-cloud
jac create .
jac start server.jac

Server runs at http://localhost:8000
Swagger docs at http://localhost:8000/docs

Frontend

cd ..   # back to project root
npm install
npm start

App runs at http://localhost:3000

---

API Endpoints

All endpoints are POST with a JSON body.

Endpoint	Description
/walker/api_setup	Initialize the warehouse graph
/walker/api_state	Get full graph state (sensors, routers, zones)
/walker/api_heartbeat	Run heartbeat scan
/walker/api_simulate_dropout	Simulate sensor going offline {"sensor_id": "SEN-042"}
/walker/api_run_pipeline	Run full 5-agent pipeline
/walker/api_diagnose	Diagnose a specific sensor {"sensor_id": "SEN-042"}
/walker/api_reset	Reset all sensors to online

---

Demo Flow

1. Open http://localhost:3000
2. Confirm ● LIVE JAC badge in the header
3. Click ▶ Simulate Dropout — SEN-042
4. Watch the agent feed fire in real time:
   • SEN-042 turns red on the map
   • Root cause diagnosed with confidence score
   • Work order generated or self-heal attempted
5. Click ↺ Reset All Sensors to restore

---

Project Structure

autonode/
├── backend/
│   ├── main.jac          # Node/edge definitions + walker agents
│   ├── server.jac        # REST API walker endpoints
│   ├── mock_data.jac     # Graph seeder (legacy)
│   ├── run.jac           # CLI demo runner
│   └── jac.toml          # Project config
├── src/
│   ├── autonode.jsx  # Main React dashboard
│   └── App.js
└── README.md

---

Why Jac?

Jac's object-spatial programming model is a natural fit for IoT mesh networks:

• The warehouse is a graph — sensors, routers, and zones are nodes
• Agents are walkers — they traverse the graph to find and fix problems
• The graph persists across API calls — state lives in the graph, not a database
• Walkers automatically become REST endpoints via jac start

This means the entire backend — data model, business logic, and API — is expressed in one language with zero boilerplate.

---

Team

Built for the Jaseci Hackathon 2026.