PHARMA FACTORY - HYBRID

Hybrid Track: Add OpenX Without Replacing Anything

You already have an industrial backbone. This path adds OpenX beside it, without disruption.

Layer OpenX alongside DeltaV, HighByte, SCADA or MES. You are not ripping and replacing; you are adding a workspace where your data is organized and read through saved queries.

What you will have

• Saved queries over your existing data
• API endpoints for LIMS, dashboards or AI tools
• Any AI over MCP, reading the same queries

All without touching your control systems.

The approach

Layer	What stays	What is added
Control	SCADA, DeltaV, PLCs	Nothing - OpenX does not touch it
Data	Historians, databases	A connection that receives a copy of the data
Query	Ad-hoc SQL, manual reports	Saved warm queries
AI	Scattered tools	Any MCP client reading the saved queries

Phase 1: Connect your data

Step 1.1: Add a connection

On the canvas

1. Open your workspace.
2. Drag a Connection node from the Node Bank onto the canvas.
3. In its inspector's Settings tab, set the Connection ID (tick Is IoT Edge Device if an IoT Edge gateway sends for it). There is no connection string to enter: the connection sets up its own device identity.
4. Save, then Deploy. Deploy registers the connection as a device on Fathym's shared IoT hub, or your own once your workspace has its IoT fabric.
5. Open the inspector's Connection Info tab. It shows the credentials your historian or gateway uses to send data (device, MQTT, AMQP and HTTP details).

Or ask Azi

Step 1.2: How your sources reach OpenX

Your historian or SCADA system forwards data to the connection's device:

Source	How it reaches OpenX
OSIsoft PI	Through an edge gateway, over MQTT
Aveva Historian	Through an edge gateway (OPC UA to MQTT)
Rockwell FactoryTalk	Over MQTT
Your own sensors	Straight to the device over MQTT, AMQP or HTTP

Phase 2: Create a monitoring surface

Step 2.1: Create the surface

On the canvas

1. Drag a Surface node from the Node Bank onto the canvas.
2. In its inspector's Settings, name it Cleanroom-Monitoring.
3. Draw a line from your connection to the surface.
4. Save. A surface is live as soon as it is saved.

Or ask Azi

Step 2.2: Check the data is flowing

Open the Live Stream panel. You should see your temperature, pressure and batch messages arriving.

Phase 3: Build compliance queries with Azi

In the main Azi panel, Azi proposes each warm query together with the surface it sits on, and it waits for a person to read and accept it (unless you have turned on Greenfield). Inside the Warm Query Manager, she drafts the query instead and you save it with the canvas Save icon.

Step 3.1: Temperature compliance

"Create a warm query that checks whether cleanroom temperature stayed between 18 and 22°C in the last 24 hours."

Data
| where EnqueuedTime > ago(24h)
| extend Payload = todynamic(RawData)
| where tostring(Payload.zone) == "Cleanroom"
| summarize MinTemp = min(todouble(Payload.temperature)),
            MaxTemp = max(todouble(Payload.temperature)),
            AvgTemp = avg(todouble(Payload.temperature))
| extend Compliant = MinTemp >= 18 and MaxTemp <= 22

Step 3.2: Batch history

"Build a query that shows every event for a batch, in time order."

Data
| extend Payload = todynamic(RawData)
| where tostring(Payload.batchId) == "112A"
| project EnqueuedTime, DeviceID, EventType = tostring(Payload.eventType), Zone = tostring(Payload.zone)
| order by EnqueuedTime asc

A warm query can declare parameters, so the batch id can be passed on each call instead of written in.

Step 3.3: Pressure threshold

"Create a query that flags any pressure reading above 1.5 bar in the mixing zone."

Data
| extend Payload = todynamic(RawData)
| where tostring(Payload.zone) == "MixingZone" and todouble(Payload.pressure) > 1.5
| project EnqueuedTime, DeviceID, Pressure = todouble(Payload.pressure)
| order by EnqueuedTime desc

Phase 4: Call your queries over REST

Every warm query is a REST endpoint.

1. Open APIs -> API Manager and its Warm Queries tab.
2. Under Configured Queries, each saved warm query has its own card, with ready-to-copy cURL, TypeScript and Python.
3. Get a key from APIs -> API Keys.

curl -X GET 'https://openx.fathym.com/ox-api/warm-queries/temp-compliance' \
  -H 'Authorization: Bearer YOUR_TOKEN'

System	Integration
LIMS	A REST call on batch completion
Dashboard	Polling the endpoint
Alerting	Polling the threshold query

Phase 5: Connect any AI over MCP

The OpenX MCP Server is at https://openx.fathym.com/ox-api/mcp. Sign in through the browser:

• Claude.ai or ChatGPT: add a custom connector, paste the URL, sign in with OpenX, pick the workspace and tick the rights to share.
• Coding agents: run fai mcp install https://openx.fathym.com/ox-api/mcp --name openx and sign in when the browser opens.

Setup for each client is in Bring Any AI. Then ask:

"Was batch 112A within its temperature limits?"

The AI reads your saved queries with call_api and answers from them. If it wants to build something new, it proposes it with manage_proposals, and a person accepts.

What you built

Component	Status
Historian connection	Built on Deploy, sending data
Cleanroom surface	Organizing your data
Compliance queries	Three saved warm queries
API endpoints	One per warm query
MCP	Any client reading the same queries

Before and after

Before	After
"The AI said it was fine"	The saved query the answer came from
Data scattered in silos	One workspace, organized by surface
AI tied to one vendor	Any MCP client reads the same queries

For what OpenX records about changes, and what it does not, see the Compliance matrix.

Next steps

If you want to...	Go to...
Try the full track	Full track
See query patterns	Query Patterns
Connect more AI clients	Bring Any AI
Learn more about Azi	Meet Azi