Introduction to Workflows

​

What is the workflow engine?

The workflow engine moves documents and business objects through configurable, auditable processes. A consultant designs a workflow once — events then start instances automatically, actions run without user involvement, and human decisions surface as tasks in users’ inboxes. A workflow is built from a small set of concepts:

• Definition — the workflow itself: a named, versioned graph of nodes and transitions
• Anchor — what the workflow operates on: either a document or a WorkView object (a Caseflow business object)
• Scope — the document types, or the application and class, the workflow is allowed to touch
• Triggers — the events that start an instance (for example, a document is created or its keywords change)
• Nodes — the steps of the process; each node can run actions when an instance arrives
• Transitions — the edges between nodes; fired automatically, after a timer, or by a user completing a task
• Actions — the side effects: update keywords, send an email, add a note, run a script, start another workflow
• Instance — one run of the workflow for one document or object
• Tasks — the human decision points; they appear in the Tasks page as buttons, optionally with a small form

​

How a workflow runs

1. A matching event arrives — the right type, inside the workflow’s scope, passing the trigger’s filter
2. The engine starts an instance at the workflow’s initial node (only one active instance per workflow per document/object)
3. The node’s on-enter actions run
4. The engine evaluates the node’s transitions: automatic transitions fire immediately when their condition passes, timer transitions fire after a delay, and task transitions wait for a user
5. The instance moves from node to node until it reaches a terminal node and completes

If an action fails, the instance routes to a designated error node — or is marked Failed and can be replayed by an administrator.

​

User surfaces

​

Who does what

​

Where to read next