AWEL Flow

Build AI workflows visually using the AWEL Flow editor — a drag-and-drop interface for composing LLM pipelines without writing code.

What is AWEL Flow?

AWEL Flow is the visual editor for AWEL (Agentic Workflow Expression Language). It lets you:

• Drag and drop operators onto a canvas
• Connect them into a DAG (Directed Acyclic Graph)
• Configure each operator's parameters
• Test and deploy workflows

Code vs Visual

AWEL workflows can be built both in Python code and visually in the Flow editor. The Flow editor generates the same underlying DAG structure.

Getting started

Step 1 — Open the Flow editor

1. Navigate to Flow in the sidebar
2. Click Create to start a new workflow

Step 2 — Add operators

The left panel shows available operators organized by category:

Category	Examples
Trigger	HTTP Trigger, Schedule Trigger
LLM	LLM Operator, Streaming LLM
RAG	Knowledge Retriever, Reranker
Agent	Agent Operator, Planning
Data	Database Query, File Reader
Transform	Text Splitter, JSON Parser
Output	Response, Stream Response

Drag an operator from the palette onto the canvas.

Step 3 — Connect operators

Click and drag from an operator's output port to another operator's input port to create a connection. Data flows along these connections.

Step 4 — Configure operators

Click on any operator to open its configuration panel. Set parameters like:

• Model name
• Prompt templates
• Database connections
• Chunk sizes
• Custom logic

Step 5 — Test and save

1. Click Run to test your workflow with sample input
2. Review the output at each stage
3. Click Save to persist the workflow

Example: Simple RAG workflow

A basic RAG workflow connects these operators:

1. HTTP Trigger — Receives the user's question
2. Knowledge Retriever — Searches the knowledge base for relevant chunks
3. Prompt Builder — Combines the question with retrieved context
4. LLM Operator — Generates the answer
5. Stream Response — Returns the streaming response

Using flows in apps

Created workflows can be used as the backend for applications:

1. Save your workflow in the Flow editor
2. Go to Apps → Create
3. Select the saved flow as the app's execution engine
4. The app inherits the flow's inputs and outputs

Managing flows

Action	How
Edit	Open a flow and modify operators/connections
Duplicate	Create a copy of an existing flow
Export	Download the flow definition as JSON
Import	Upload a flow definition JSON file
Delete	Remove a flow from the list

Installing community operators

Community operators from the dbgpts repository automatically appear in the Flow editor after installation:

dbgpts install <operator-package>

Next steps

Topic	Link
AWEL concepts	AWEL
AWEL Python tutorial	AWEL Tutorial
AWEL cookbook	AWEL Cookbook
Community operators	dbgpts