Workflow Builder

The Workflow Builder is a visual, no-code automation platform that lets you create, test, and deploy automated workflows connecting AI, external services, and custom logic.

What is a Workflow?

A workflow is an automated sequence of steps that runs when triggered. Each step is a node on a visual canvas, connected by edges that define the execution flow.

Creating a Workflow

Search for the latest AI news, summarize the results,
and send a Slack message to #technology with the summary.

Available Nodes

AI Nodes

Web and Data

Logic and Control

Communication

Integrations

AI-Powered Generation

You can create or modify workflows using natural language:

1. Click the AI prompt input in the editor
2. Describe what you want in plain English
3. Choose the mode:
   • Create — Generate a new workflow from scratch
   • Modify — Update the existing workflow
   • Append — Add new steps to the end
4. Review the generated nodes and connections
5. Customize as needed

The AI respects your configured AI provider and model preferences.

Integrations

Configure credentials for external services under Workflows → Integrations.

Managing Credentials

• All credentials are encrypted before storage
• Credentials are scoped to your account or organization
• Use Test Connection to verify credentials before saving
• Credentials are masked in the UI for security

Publishing and Versioning

Publishing a Workflow

When you publish a workflow:

1. The current state is saved as a numbered version
2. You can add a changelog describing the changes
3. A webhook URL is generated (if webhook trigger is enabled)
4. The workflow becomes active for scheduled triggers

Version History

• Every publish creates a new version
• View all versions in the Version History panel
• Rollback to any previous version instantly
• Compare published vs. current draft changes

Webhook Triggers

Published workflows with webhook triggers get a unique URL:

POST /api/workflows/trigger/{workflowId}

Include the webhook secret in the request header for authentication. Pass input data as JSON in the request body.

Execution and Monitoring

Running Workflows

Workflows can be triggered:

• Manually — Click "Run" in the editor
• Via Webhook — Send an HTTP POST request
• On Schedule — Automatically at configured times
• By Event — Triggered by system events

Execution Panel

The execution panel shows real-time progress:

• Per-node logs with input data, output data, and timing
• Status indicators — Pending, Running, Completed, Failed
• Error messages for debugging failed nodes
• Duration tracking for performance analysis

Execution History

View past executions in the Run History tab:

• Filter by status (Completed, Failed, Cancelled, Timed Out)
• Inspect individual node results
• Copy execution data for debugging
• View trigger input data

Human-in-the-Loop Approvals

Workflows can pause for manual approval before proceeding:

1. Add an approval step to your workflow
2. When execution reaches this step, it pauses
3. Approvers receive a notification
4. Review the context and Approve or Reject
5. Optionally add feedback or comments
6. Workflow continues or stops based on the decision

Workflow Statuses

Best Practices

Design

• Start simple — Begin with 2-3 nodes, test, then expand
• Use conditions — Add branching for error handling
• Name nodes clearly — Descriptive names help with debugging

Testing

• Test manually first — Always run manually before publishing
• Check node outputs — Verify each node produces expected results
• Use test data — Don't use production credentials during testing

Production

• Monitor executions — Check the run history regularly
• Set up error handling — Use conditions to handle failures gracefully
• Version your changes — Publish with meaningful changelogs
• Rotate credentials — Update integration credentials periodically

Next Steps