How to Connect OPC-UA (SCADA/DCS) to an AI Agent: A Factory Automation Case Study

The Challenge of Industrial Data Integration

In a typical factory, hundreds of sensors stream data into an OPC-UA server – temperature, pressure, flow rates, vibration levels. Engineers need real-time visibility and automatic reactions when values go out of range. Traditional solutions require custom scripts, dashboards, and complex alerting rules. What if you could just ask an AI agent to handle it?

ASI Biont is an AI agent that connects directly to industrial controllers and servers. It doesn't need a dashboard or a plugin – everything happens through a chat conversation. Describe your OPC-UA endpoint and what you want to monitor, and the AI writes the integration code on the fly.

How ASI Biont Connects to OPC-UA

ASI Biont uses the open‑source library opcua‑asyncio (asyncua) for OPC UA client connections. The AI leverages the industrial_command tool with two primary commands:

• read_variable – read a node's value
• write_variable – write a new value to a node

Example command sent from the chat:

industrial_command(
  protocol='opcua',
  command='read_variable',
  endpoint='opc.tcp://192.168.1.100:4840',
  node_id='ns=2;s=Temperature_Sensor_01'
)

Behind the scenes, the AI establishes a secure connection, reads the variable, and returns the value – all in under a second. If more complex logic is needed (e.g., reading multiple tags or applying a math model), the AI can also generate a Python script that runs in the execute_python sandbox with the same opcua‑asyncio library.

Real-World Case: Monitoring a Chemical Reactor

Problem

A chemical plant has an OPC-UA server aggregating data from 30 reactors. Operators want automatic alerts when a reactor's temperature exceeds 85°C or pressure goes above 12 bar. They also need a daily trend report.

Solution with ASI Biont

The engineer simply tells the AI: "Connect to our OPC-UA server at opc.tcp://plant-opc.local:4840, every 10 seconds read tags 'Reactor_Temp' and 'Reactor_Pressure', and if temperature > 85°C send a Telegram alert to our group. Also save hourly averages to a file."

The AI generates a Python script that runs in the cloud sandbox (using asyncua) and executes it. The script loops (with a 30‑second limit per run, but ASI Biont can schedule periodic execution via cron). The code looks like this (simplified):

import asyncio
from asyncua import Client

async def monitor():
    async with Client(url='opc.tcp://plant-opc.local:4840') as client:
        temp_node = await client.nodes.root.get_child(['0:Objects', '2:Reactor_Temp'])
        press_node = await client.nodes.root.get_child(['0:Objects', '2:Reactor_Pressure'])
        temp = await temp_node.read_value()
        press = await press_node.read_value()
        if temp > 85:
            # send Telegram alert (using requests or aiohttp)
            pass

asyncio.run(monitor())

No manual wiring – the AI handles the OPC UA discovery, security (anonymous or username/password), and error handling. The user sees the result in the chat.

Results

• Within minutes, the plant had automated alerts to a Telegram chat.
• The AI also created a daily summary (using xlsxwriter to write an Excel file).
• When a pressure spike was detected, the AI wrote a new setpoint to the controller's Pressure_Setpoint node using write_variable, bringing the system back to safe limits.

Why This Approach Matters

Getting Started

You don't need to install any software on the OPC-UA server. Just provide the endpoint and credentials (if any) in the ASI Biont chat. The AI agent uses opcua-asyncio inside its sandbox – no extra downloads. For local machines behind firewalls, ASI Biont's Hardware Bridge can tunnel the connection if needed, but for pure OPC-UA over TCP, the cloud sandbox works perfectly if the server is reachable.

Try it yourself: tell the AI "Connect to my OPC-UA server at opc.tcp://192.168.0.10:4840, read all analog sensor values, and create a real-time dashboard in the chat using markdown tables." It will do just that.

Conclusion

OPC-UA is the backbone of Industry 4.0, but extracting intelligence from it often requires heavy custom code. ASI Biont removes that barrier. By describing your requirements in natural language, you get a fully automated integration – from data collection to smart alerts and control. No more waiting for developers; the AI becomes your industrial IoT engineer.

Visit asibiont.com and start connecting your SCADA, DCS, or any OPC-UA device to an AI agent today.