Workflows
Visual BuilderBuild complex AI pipelines with a visual node-based editor. Chain models, add logic, integrate tools, and deploy workflows with a single click.
Visual Workflow Builder
Drag-and-drop interface for building AI pipelines without code.
Key Features
Visual Editor
Drag and drop nodes, connect them visually, and see your pipeline at a glance. No coding required.
Real-time Execution
Watch your workflow execute step-by-step with live output streaming and debugging.
Conditional Logic
Add branches, loops, and conditions to create dynamic AI applications.
API & Webhooks
Deploy as an API endpoint or trigger from external webhooks.
RAG Integration
Connect to your RAG instances for knowledge-grounded responses.
Template Library
Start with pre-built templates for common use cases like chatbots, data extraction, and more.
How It Works
Create a Workflow
Start from scratch or pick a template. Each workflow belongs to an agent and can have multiple versions.
Add Nodes
Drag nodes from the palette: LLM calls, prompts, code blocks, HTTP requests, conditionals, and more.
Connect Nodes
Draw connections between nodes to define data flow. Outputs from one node become inputs to the next.
Test & Deploy
Run your workflow in the editor, then deploy it as an API or attach triggers like webhooks or schedules.
Example: Customer Support Bot
User message
Find docs
Generate reply
Needs escalation?
Response
This workflow receives customer messages, searches your knowledge base for relevant documentation, generates a response with context, and optionally escalates complex issues.
Available Node Types
| Category | Nodes |
|---|---|
| Flow Control | Start, Output, Condition, Loop, Parallel, Merge |
| AI Models | LLM, Image Generation, Embeddings, Vision |
| Data | Prompt Template, RAG Query, JSON Transform, Variable |
| Integrations | HTTP Request, Webhook, Database Query |
| Code | JavaScript, Python (sandboxed execution) |
See Node Types documentation for detailed information on each node.
API Access
Every workflow can be executed via API:
Supports streaming responses via SSE for real-time output. See the Triggers documentation for webhooks and scheduled execution.
Pricing
Workflow execution is billed based on the AI models used within the workflow. Building and testing in the editor is free—you only pay for actual LLM/embedding/image generation calls when executing.