-success){kind=icon}
Sample process: use AI + Gline modules (REST API, oBIX, FTP Server) to automate Niagara data point collection and web dashboard creation. A reference for integration projects.
Key enabler: The Gline REST API can configure Modbus client proxyExt addresses (dataAddress, regType, dataType) remotely. This means AI can create a Modbus point AND set its register address in one automated flow — no manual Workbench steps required for Modbus point configuration.
This repository demonstrates a sample AI automation process for Niagara 4 using Gline's free modules. It shows how AI can:
- Auto-discover Modbus devices and collect data points
- Create and configure points via REST API
- Generate PX/HTML web pages automatically
- Deploy and verify pages via FTP Server (auto-login via SCRAM, verify bindings, fix errors)
The entire workflow runs on DeepSeek V4 Flash + OpenClaw and costs only cents in AI tokens.
| API | Port | Auth | Cost |
|---|---|---|---|
| oBIX | 80/443 | SCRAM-SHA-256 / HTTP Basic | 🆓 Free (built-in) |
| Gline REST API | Configurable (default 8081) | None (LAN) | 🆓 Free (no SMA required) |
The AI (tested on DeepSeek) acts as a Niagara engineer — it reads the station structure, decides what operations to perform, then executes them via REST API and oBIX calls. No human in the loop.
AI scans a Modbus device (or any protocol), reads the register map, and creates all necessary points automatically:
- AI sends
LIST /Drivers/via REST API to understand current station structure - AI determines the correct point types (NumericWritable, BooleanWritable, etc.)
- AI batch-creates points via
ADDoperations - AI configures each point's Modbus register address, data type, and function code via oBIX
PUT
After creating points, AI binds them to PX visualization pages:
- AI generates the PX XML template with proper bindings
- AI writes the XML to station via
StringToFilemodule - AI establishes
Linkconnections between points and UI components - Result: a fully functional web dashboard — zero manual binding
AI can deploy new modules and services to the station:
- Copy JAR to modules directory
- AI creates the required service component via
ADD - AI configures service parameters via
MOD - AI verifies the service is running via
READ
AI reads alarms and historical data on demand:
- Alarms:
alarm:|bql:select ... where alarmClass = 'Red_High' - History:
history:|slot:/History/xxx|bql:select ... where timestamp > '...' - Real-time points:
station:|slot:/Drivers/...|bql:select name,out.value,out.status - Results returned as structured JSON
┌──────────────────────────────────────────────────────┐
│ AI (DeepSeek) │
│ Understands Niagara structure, decides operations │
└──────────┬───────────────────────────────┬───────────┘
│ │
REST API (8081) oBIX (80/443)
┌──────────┐ ┌──────────────┐
│ LIST │ │ GET /out │
│ ADD │ │ PUT /address │
│ REMOVE │ │ PUT /regType │
│ RENAME │ │ Link create │
│ DUPLICATE│ │ Alarm query │
│ LINK │ │ History query│
│ MOD │ └──────────────┘
│ READ │
└──────────┘
│ │
└───────────┬───────────────────┘
▼
┌─────────────────────┐
│ Niagara Station │
│ - Points created │
│ - Links established │
│ - PX pages deployed │
│ - Alarms monitored │
└─────────────────────┘
| Step | Operation | API | Description |
|---|---|---|---|
| 1 | LIST /Drivers |
REST | AI scans existing drivers and services |
| 2 | ADD ModbusTcpNetwork |
REST | Creates Modbus TCP network if not present |
| 3 | MOD ipAddress=192.168.2.140 |
REST | Configures device IP address |
| 4 | CP template→6 points |
REST | Copies 6 point templates from a pre-configured source |
| 5 | RENAME each point |
REST | Names: Set_Temp, Zone_Temp, Fan_Speed, On_Off, Filter_Status, Supply_Air_T |
| 6 | PUT proxyExt/address |
oBIX | Sets Modbus register addresses for each point |
| 7 | PUT proxyExt/regType |
oBIX | Sets holding/input/coil/discrete for each point |
| 8 | GET /out |
oBIX | Verifies all points are reading live values |
| 9 | Generate PX XML | AI | AI creates a PX page with bound labels |
| 10 | StringToFile write PX |
REST | Deploys the PX page to the station |
| 11 | Browser open | Web | Real-time dashboard is live |
This is our flagship AI automation demonstration — a fully autonomous integration of a Johnson Controls T8600 Modbus thermostat into a Niagara 4.14 station.
| Tool | Role |
|---|---|
| DeepSeek V4 Flash | AI reasoning engine |
| OpenClaw | AI agent framework |
| Gline REST API | Component lifecycle (ADD, MOD, CP, RENAME, LINK) |
| oBIX | Configure proxyExt (register address, type, function code) |
| Gline FTP Server | Deploy generated PX/HTML pages to station |
Step 1 User tells AI: "Integrate JCI T8600 thermostat at 192.168.x.x"
↓
Step 2 AI searches the internet for T8600 Modbus register map
(discovers holding/input/coil/discrete registers automatically)
↓
Step 3 AI connects to the Niagara station via REST API
↓
Step 4 AI scans existing Modbus network - finds the thermostat device
↓
Step 5 AI adds a new Device under the Modbus network via REST API ADD
↓
Step 6 AI creates ModbusClient points with correct register addresses
(Set_Temp, Zone_Temp, Fan_Speed, On_Off, Filter_Status, Supply_Air_T)
↓
Step 7 AI configures each point's proxyExt via oBIX PUT
(dataAddress, regType, dataType)
↓
Step 8 AI names all points properly via REST API RENAME
↓
Step 9 AI generates a PX dashboard (or BajaScript HTML page)
with bound labels showing all 6 point values
↓
Step 10 AI uploads the page to station via FTP Server
↓
Step 11 AI authenticates via SCRAM (same as a human logging into the web UI),
inspects the PX/HTML file, checks bindings, verifies data displays correctly
↓
Step 12 If errors found, AI generates a corrected page and re-uploads via FTP Server
↓
✅ Complete! Real-time dashboard live, remote writable
| Before AI Automation | After AI Automation |
|---|---|
| Manual Modbus register mapping | AI searches spec sheets automatically |
| Manual point creation (drag-drop in Workbench) | AI creates via REST API in milliseconds |
| Manual proxyExt configuration | AI configures via oBIX in seconds |
| Manual PX binding and UI design | AI generates PX/HTML pages autonomously |
| Manual FTP upload and testing | AI uploads, verifies, and fixes errors |
| Hours to days per device | Minutes per device |
| $500-$2,000+ in engineering labor | $0.03-0.05 in AI token cost |
On DeepSeek V4 Flash, the entire T8600 integration consumes approximately a few cents (USD) in AI tokens. The same work traditionally requires hours of a Niagara engineer's time.
The bottom line: Point binding, UI creation, and Modbus device integration are no longer time-consuming tasks. AI handles the entire pipeline — from register discovery to live dashboard deployment.
| Service | Price | Notes |
|---|---|---|
| AI automation pipeline (current capabilities) | Free | Tested and working on DeepSeek |
| Custom AI automation development | Contact us | Tailored to your specific station and workflow |
The current capability is offered free of charge — we want to demonstrate what's possible. If you see value and want us to build a custom automation pipeline for your Niagara system, we'd be happy to discuss.
Need a custom REST API for your AI workflow? The free Gline REST API covers common operations. If you need custom endpoints, advanced security, or specialized data pipelines for AI automation — we can build it. Contact us for a discussion.
| Component | Requirement |
|---|---|
| Niagara | 4.14 or later |
| Gline REST API | Free module — get it here |
| oBIX | Built-in (enabled by default) |
| AI Model | DeepSeek (tested) or any OpenAI-compatible model |
| Network | AI host must reach station REST API and oBIX ports |
If you're interested in AI-powered Niagara automation for your projects:
- We can build custom pipelines for your specific use case
- We can deploy the AI agent inside your network
- We can integrate with your existing tools and workflows
Contact: jason.zhang@gline-net.com | WhatsApp: +86 1380 190 9968
© 2026 Shanghai Gline Net Co., Ltd. All rights reserved.