Execution Engine
LyznFlow’s execution engine is the heart of the platform. It takes your visual workflow and turns it into actual running code.How Execution Works
Every time a workflow runs, it creates an Execution. An execution is a single instance of your workflow running from start to finish.Execution Lifecycle
- Trigger: An event (like a chat message or a scheduled time) initiates the process.
- Queuing: The execution is placed in a queue (
Pendingstate). - Running: The engine picks up the job and starts executing nodes one by one (
Runningstate). - Completion:
- Success: All nodes finished correctly.
- Error: A node failed, and the workflow stopped (unless error handling was configured).
Watch Mode
Watch Mode is a powerful feature for developers. It allows you to “watch” the backend execution engine in real-time directly on your canvas.How it Works
When you toggle Watch Mode ON:- The editor subscribes to a real-time WebSocket channel for this workflow.
- As the backend engine executes each node, it broadcasts events (
nodeStarted,nodeFinished,nodeError). - The editor receives these events and animates the nodes:
- Pulse Green: Node is currently running.
- Solid Green Border: Node finished successfully.
- Red Border: Node failed.
- Output Preview: You can click a finished node to see exactly what output it produced.
Note: Watch Mode only affects your view of the editor. It does not change how the workflow runs on the server.
Publishing Your Agent
Once your workflow is working perfectly in Watch Mode, you are ready to publish.Public vs Private
- Private (Default): Only you (and your team) can trigger or chat with this agent. Perfect for development and internal tools.
- Public: The agent is listed in the Lyzn Marketplace or accessible via a public link. Anyone can use it.
Versioning
LyznFlow handles versioning automatically.- Draft: The version currently on your canvas.
- Published: The version that is live for users.
- When you click Publish, your current Draft becomes the new Published version.
Next Steps: Learn how to build advanced AI Agents.

