You operate SwiftMQ. Your routers move messages all day: machine telemetry, orders, status events, sensor readings. Flow Director runs alongside SwiftMQ. Yet most of what flows through your queues and topics is never looked at. The Flow Director MCP Server closes that gap: you describe what you want to see, and an AI builds the Flow Director app for you.
Your data is already there
As a SwiftMQ operator you see the plumbing: queues, topics, message rates, backlogs, connections. What you usually don't see is what the messages mean. Which machine is overheating? How many orders are stuck? Which station is the bottleneck right now?
That information is in the messages you already route. It has value – but only if someone makes it visible.
Building the app is the hard part
Flow Director can turn those messages into live dashboards, alerts and KPIs without code. But someone still has to build the app: design the flows, wire the components, parse the messages, set up the dashboards and test everything. Even without code, that takes time and Flow Director know-how. So it often doesn't happen – or only for the one use case that justified a project.
Describe it – the AI builds it
With the Flow Director MCP Server you connect an AI client such as Claude Code or Codex to an empty Flow Director app and describe what you need in plain language:
-
what the messages contain and where they come from (topics, queues, REST),
-
what you want to see (status, trends, counters),
-
when you want to be alerted.
The AI asks what it needs to know, proposes a design and builds the app after you confirm: flows, dashboard, menu entry. Then it starts the app and checks that everything works. A complete app takes about 30 to 45 minutes.
What you get
-
Visibility into your message data – live dashboards and alerts on what flows through SwiftMQ, not just on queue depths.
-
Apps in under an hour – a complete app in 30 to 45 minutes, without having to know Flow Director inside out.
-
Unlimited apps – one license unlocks the MCP server; every app you build with it is license-free.
-
More questions answered – small, specific use cases become worth doing, not only the big project.
-
A normal Flow Director app – the result is an ordinary app you can open, change and extend by hand or with a follow-up prompt.
-
Your data stays in your system – the app runs in your Flow Director on your SwiftMQ.
You stay in control
-
Same rights as you: the AI works through the Flow Director user interface. It can see and do exactly what the logged-in user can – nothing more.
-
Admins only, own apps only: AI sessions are available to admin users and only for apps you created yourself.
-
Design first: the AI proposes a design and waits for your confirmation before it builds anything.
-
Watch and stop: you see every step live in your browser and can end the session at any time.
-
Full log: every action is logged and can be downloaded at the end of the session.
No AI in production
The usual workflow keeps the AI out of production entirely:
-
Build on a development system. The AI session runs against a Flow Director instance on dev, connected to a dev SwiftMQ.
-
Check the app into Git. Once the app does what you want, commit it to your Git repository – reviewable and versioned like any other change.
-
Deploy to production from Git. Production pulls the app from Git and runs it. No MCP server, no AI session, no AI access.
The AI helps you build faster. What runs in production is a normal, reviewed Flow Director app.
What you need
-
Flow Director running with your SwiftMQ
-
One Flow Director App license to unlock the MCP server in the System app (a 30-day trial is available). With it you can build any number of apps – the apps themselves need no license.
-
An admin user
-
Claude Code or Codex on your computer
Get started
Follow the step-by-step guide: create an empty app, pair your AI client, describe the app.
Build an App with AI: Empty App, Pair, Prompt