Circuitry Logo
Circuitry
Open Editor

Building Workflows

Learn how to create, manage, and optimize workflows in Circuitry.

Workflow Basics

A workflow is a sequence of connected nodes that process data from start to finish. Workflows can be linear, branched, or parallel, depending on your automation needs.

Creating a New Workflow

Starting from Scratch

  1. Open the Circuitry Editor
  2. Your canvas starts with a Start Node already in place - this is your workflow's entry point
  3. Add additional nodes by dragging from the sidebar
  4. Connect nodes to define data flow
  5. Configure each node's settings
  6. Save your workflow with a descriptive name

Using Templates

Coming soon: Pre-built workflow templates for common use cases.

Workflow Components

Canvas Controls

  • Pan: Click and drag on empty canvas space
  • Zoom: Use mouse wheel or trackpad
  • Fit View: Double-click empty space
  • Multi-select: Hold Shift and click nodes
  • Delete: Select node(s) and press Delete key

Connection Rules

  • One-to-One: Most nodes have single input/output
  • One-to-Many: Fork nodes split to multiple outputs
  • Many-to-One: Join nodes merge multiple inputs
  • Type Safety: Connections validate data compatibility

Execution Flow

Sequential Execution

By default, nodes execute in sequence:

Start → Node A → Node B → Node C → End

Each node waits for the previous to complete before starting.

Parallel Execution

Use Fork and Join nodes for parallel processing:

       ┌→ Branch A →┐
Start →→ Fork → Branch B → Join → End
       └→ Branch C →┘

All branches execute simultaneously, improving performance.

Conditional Execution

Use Condition nodes to create branching logic:

Start → Condition → [if true] → Process A → End
                 └→ [if false] → Process B → End

Data Flow

Understanding Data Passing

  1. Each node receives input from connected predecessor(s)
  2. Processes the input according to its configuration
  3. Produces output for successor node(s)
  4. Data is passed as JSON between nodes

Data Transformation

Nodes can transform data in various ways:

  • Agent Nodes: Generate new content based on prompts
  • Action Nodes: Modify, filter, or enrich data
  • Loop Nodes: Process arrays element by element
  • Join Nodes: Combine multiple data streams

Variable References

Reference data from previous nodes using template syntax:

{{nodeName.output.field}}
{{input.propertyName}}
{{previousNode.result}}

Workflow Management

Saving Workflows

Workflows auto-save as you work. Manual save options:

  • Ctrl/Cmd + S: Quick save
  • Save As: Create a copy with new name
  • Version History: Track changes over time

Organizing Workflows

Best practices for workflow organization:

  1. Descriptive Names: Use clear, action-oriented names
  2. Folders: Group related workflows (coming soon)
  3. Tags: Add tags for easy filtering
  4. Documentation: Add descriptions to complex workflows

Sharing Workflows

Options for collaboration:

  • Export: Download workflow as JSON
  • Import: Load workflow from JSON file
  • Share Link: Generate shareable URL (coming soon)
  • Team Workspaces: Collaborate with team members

Execution Monitoring

Real-time Feedback

During execution, observe:

  • Node States: Color-coded execution status
  • Progress Indicators: Loading spinners on active nodes
  • Data Preview: Hover nodes to see input/output
  • Execution Time: Duration for each node
  • Error Messages: Detailed error information

Execution Logs

Access detailed logs showing:

  • Timestamp for each operation
  • Input/output data for debugging
  • Error stack traces
  • Performance metrics

Performance Optimization

Best Practices

  1. Minimize API Calls: Batch operations when possible
  2. Use Caching: Enable caching for repeated operations
  3. Parallel Processing: Use Fork/Join for independent tasks
  4. Efficient Prompts: Keep AI prompts concise and focused
  5. Data Filtering: Process only necessary data

Resource Limits

Be aware of system limits:

  • Execution Timeout: 5 minutes per workflow
  • Memory Limit: 512MB per execution
  • API Rate Limits: Varies by service
  • Parallel Branches: Maximum 10 concurrent

Error Handling

Common Issues

  1. Connection Errors: Ensure nodes are properly connected
  2. Configuration Errors: Validate node settings
  3. Data Format Errors: Check JSON structure
  4. API Failures: Handle rate limits and timeouts
  5. Model Errors: Verify AI model availability

Debugging Strategies

  1. Test Incrementally: Execute partial workflows
  2. Use Sample Data: Test with known inputs
  3. Check Logs: Review execution logs for details
  4. Add Logging: Use Action nodes to log intermediate data
  5. Error Boundaries: Use conditions to handle failures

Advanced Techniques

Nested Workflows

Call one workflow from another (coming soon):

Main Workflow → Execute Sub-Workflow → Continue

Dynamic Node Configuration

Configure nodes based on runtime data:

// In Action node
const config = {
  model: input.preferredModel || 'gpt-5',
  temperature: input.creativity || 0.7
}

Custom Functions

Write JavaScript in Action nodes:

// Data transformation
const processed = input.items.map(item => ({
  ...item,
  timestamp: Date.now(),
  status: 'processed'
}));

return { processed };

Next Steps