Workflows

Build multi-step agent pipelines with a visual drag-and-drop DAG builder. Chain LLM calls, tool invocations, and conditional logic into automated workflows.

What Are Workflows?

A workflow is a directed acyclic graph (DAG) of steps that execute in order. Each step can be an LLM call, tool invocation, condition check, or custom logic. Workflows let you build complex automations without code.

Node Types

Node	Description
Start	Entry point. Defines input variables.
LLM	Calls an AI model with a prompt. Output goes to next node.
Tool	Invokes a registered tool with parameters.
Condition	Branches based on a condition (if/else).
End	Exit point. Returns the final output.

Create a Workflow (API)

Create Workflowbash

curl -X POST http://localhost:8000/api/v1/workflows \
  -H "Authorization: Bearer YOUR_API_KEY" \
  -d '{
    "name": "support-pipeline",
    "description": "Classify then respond to support tickets",
    "nodes": [
      { "id": "start", "type": "start" },
      { "id": "classify", "type": "llm", "config": { "prompt": "Classify this ticket: {{input}}", "model": "gpt-4o-mini" }},
      { "id": "respond", "type": "llm", "config": { "prompt": "Write a response for this {{classify.output}} ticket: {{input}}", "model": "gpt-4o" }},
      { "id": "end", "type": "end" }
    ],
    "edges": [
      { "from": "start", "to": "classify" },
      { "from": "classify", "to": "respond" },
      { "from": "respond", "to": "end" }
    ]
  }'

Visual Builder

Use the Workflows tab in the dashboard for a visual drag-and-drop builder. Much easier than writing JSON manually!

Knowledge Base

Chat Widget