What Are Config Workflows?
Each Config Workflow is a versioned, publishable flow that defines:- How incoming requests are routed to different AI providers
- What plugins run before or after model calls (e.g., PII detection, content moderation)
- What branching logic determines the next step based on model output
- How fallbacks and retries are handled when providers fail
Think of Config Workflows as programmable middleware for your AI requests — defined visually, versioned automatically, and deployed without code changes.
Example: Email Job Listings
Here’s a real Config Workflow that processes job application emails: This workflow:- fetch_gmail (Plugin) — Fetches incoming emails
- cv_scorer (Router) — Routes the email to an AI model for CV scoring
- score_route (Branch) — Branches based on the score:
score > 7→ Send intro invite + check Calendly slots + update calendarscore < 7→ Put on holdno CV→ Request missing documents
- Various Plugin nodes execute the final actions (send mail, update Calendly, etc.)
Node Types
Config Workflows support several node types. See the Nodes Reference for full details.Router
Routes requests to AI models. Configure the provider, model, and prompts.
Branch
Conditional logic with IF / ELSE IF / DEFAULT paths based on expressions evaluated against node outputs.
Plugin
Execute actions like PII detection, content moderation, API calls, email sending, and more.
Getting Started
Design Your Flow
Add nodes to the canvas by clicking the + button. Connect nodes by dragging edges between them.
Configure Nodes
Click any node to open its settings panel. Set routing targets, branch conditions, or plugin configurations.
Nodes Reference →
Learn about all available node types — Router, Branch, and Plugin.

